www.researchgate.net
Open in
urlscan Pro
2606:4700:4400::6812:2909
Public Scan
URL:
https://www.researchgate.net/publication/320453789_A_generalized_simulation_development_approach_for_predicting_refugee_desti...
Submission: On December 02 via api from US — Scanned from DE
Submission: On December 02 via api from US — Scanned from DE
Form analysis 3 forms found in the DOM
GET search
<form method="GET" action="search" class="lite-page__header-search-input-wrapper"><input type="hidden" name="context" readonly="" value="publicSearchHeader"><input placeholder="Search for publications, researchers, or questions" name="q"
autocomplete="off" class="lite-page__header-search-input"><button
class="nova-legacy-c-button nova-legacy-c-button--align-center nova-legacy-c-button--radius-full nova-legacy-c-button--size-s nova-legacy-c-button--color-white nova-legacy-c-button--theme-ghost nova-legacy-c-button--width-square lite-page__header-search-button"
type="submit" width="square"><span class="nova-legacy-c-button__label"><svg aria-hidden="true"
class="nova-legacy-e-icon nova-legacy-e-icon--size-s nova-legacy-e-icon--theme-bare nova-legacy-e-icon--color-inherit nova-legacy-e-icon--luminosity-medium">
<use xlink:href="/m/4154823318689881/images/icons/nova/icon-stack-s.svg#magnifier-s"></use>
</svg></span></button></form>Name: loginForm — POST https://www.researchgate.net/login?_sg=F0oU4A_NOPT5xThaiK3zFkgFyQBIirKNjULLyLz-CuXRvUpAeAb2o1LDRB4wfE0rxCRfC6AmJuLPHQ
<form method="post" action="https://www.researchgate.net/login?_sg=F0oU4A_NOPT5xThaiK3zFkgFyQBIirKNjULLyLz-CuXRvUpAeAb2o1LDRB4wfE0rxCRfC6AmJuLPHQ" name="loginForm" id="headerLoginForm"><input type="hidden" name="request_token"
value="aad-GRhc42RQ2jSQY1Vc6xkYgVNXZQsGi8hGLKNfp6Xd74fkx5Yup5lRNdwLl24GUdi2lw/MVRmS8vfOY6bPSVOjTLUHl6t7hACiyN+zOC5ENzuLZFVHXb+38t2/5HGoISCrJm98ZYzStnwj3VfH1tfP8ANGeIDk7V4xJyXtzoChldQS0OE+xG1g+uTIyAm5wVyfgiy9tMUO6kmvQkfbY3nJnz4EKsX75hJMkFVXvXEP4JUEmw08zn30j4n3UvJIRYX6bcxyAFcmwF35LaaHl9w="><input
type="hidden" name="urlAfterLogin" value="publication/320453789_A_generalized_simulation_development_approach_for_predicting_refugee_destinations"><input type="hidden" name="invalidPasswordCount" value="0"><input type="hidden" name="headerLogin"
value="yes">
<div class="lite-page__header-login-item"><label class="nova-legacy-e-text nova-legacy-e-text--size-m nova-legacy-e-text--family-sans-serif nova-legacy-e-text--spacing-none nova-legacy-e-text--color-inherit lite-page__header-login-label"
for="input-header-login">Email <div class="lite-page-tooltip "><svg aria-hidden="true" class="nova-legacy-e-icon nova-legacy-e-icon--size-s nova-legacy-e-icon--theme-bare nova-legacy-e-icon--color-inherit nova-legacy-e-icon--luminosity-medium">
<use xlink:href="/m/4154823318689881/images/icons/nova/icon-stack-s.svg#info-circle-s"></use>
</svg>
<div class="lite-page-tooltip__content lite-page-tooltip__content--above">
<div class="nova-legacy-e-text nova-legacy-e-text--size-s nova-legacy-e-text--family-sans-serif nova-legacy-e-text--spacing-none nova-legacy-e-text--color-inherit"><b>Tip:</b> Most researchers use their institutional email address as their
ResearchGate login</div>
<div class="lite-page-tooltip__arrow lite-page-tooltip__arrow--above">
<div class="lite-page-tooltip__arrow-tip"></div>
</div>
</div>
</div></label></div><input type="email" required="" placeholder="" id="input-header-login" name="login" autocomplete="email" tabindex="1"
class="nova-legacy-e-input__field nova-legacy-e-input__field--size-m lite-page__header-login-item nova-legacy-e-input__ambient nova-legacy-e-input__ambient--theme-default">
<div class="lite-page__header-login-item"><label class="lite-page__header-login-label"
for="input-header-password">Password</label><a class="nova-legacy-e-link nova-legacy-e-link--color-blue nova-legacy-e-link--theme-bare lite-page__header-login-forgot" href="application.LostPassword.html">Forgot password?</a></div><input
type="password" required="" placeholder="" id="input-header-password" name="password" autocomplete="current-password" tabindex="2"
class="nova-legacy-e-input__field nova-legacy-e-input__field--size-m lite-page__header-login-item nova-legacy-e-input__ambient nova-legacy-e-input__ambient--theme-default"><label
class="nova-legacy-e-checkbox lite-page__header-login-checkbox"><input type="checkbox" class="nova-legacy-e-checkbox__input" aria-invalid="false" name="setLoginCookie" tabindex="3" value="yes" checked=""><span
class="nova-legacy-e-checkbox__checkmark"></span><span class="nova-legacy-e-checkbox__label"> Keep me logged in</span></label>
<div
class="nova-legacy-l-flex__item nova-legacy-l-flex nova-legacy-l-flex--gutter-m nova-legacy-l-flex--direction-column@s-up nova-legacy-l-flex--align-items-stretch@s-up nova-legacy-l-flex--justify-content-center@s-up nova-legacy-l-flex--wrap-nowrap@s-up">
<div class="nova-legacy-l-flex__item"><button
class="nova-legacy-c-button nova-legacy-c-button--align-center nova-legacy-c-button--radius-m nova-legacy-c-button--size-m nova-legacy-c-button--color-blue nova-legacy-c-button--theme-solid nova-legacy-c-button--width-full" type="submit"
width="full" tabindex="4"><span class="nova-legacy-c-button__label">Log in</span></button></div>
<div class="nova-legacy-l-flex__item nova-legacy-l-flex__item--align-self-center@s-up">
<div class="nova-legacy-e-text nova-legacy-e-text--size-s nova-legacy-e-text--family-sans-serif nova-legacy-e-text--spacing-none nova-legacy-e-text--color-inherit">or</div>
</div>
<div class="nova-legacy-l-flex__item">
<div
class="nova-legacy-l-flex__item nova-legacy-l-flex nova-legacy-l-flex--gutter-m nova-legacy-l-flex--direction-column@s-up nova-legacy-l-flex--align-items-center@s-up nova-legacy-l-flex--justify-content-flex-start@s-up nova-legacy-l-flex--wrap-nowrap@s-up">
<div class="nova-legacy-l-flex__item">
<a href="connector/google"><div style="display:inline-block;width:247px;height:40px;text-align:left;border-radius:2px;white-space:nowrap;color:#444;background:#4285F4"><span style="margin:1px 0 0 1px;display:inline-block;vertical-align:middle;width:38px;height:38px;background:url('images/socialNetworks/logos-official-2019-05/google-logo.svg') transparent 50% no-repeat"></span><span style="color:#FFF;display:inline-block;vertical-align:middle;padding-left:15px;padding-right:42px;font-size:16px;font-family:Roboto, sans-serif">Continue with Google</span></div></a>
</div>
</div>
</div>
</div>
</form>Name: loginForm — POST https://www.researchgate.net/login?_sg=F0oU4A_NOPT5xThaiK3zFkgFyQBIirKNjULLyLz-CuXRvUpAeAb2o1LDRB4wfE0rxCRfC6AmJuLPHQ
<form method="post" action="https://www.researchgate.net/login?_sg=F0oU4A_NOPT5xThaiK3zFkgFyQBIirKNjULLyLz-CuXRvUpAeAb2o1LDRB4wfE0rxCRfC6AmJuLPHQ" name="loginForm" id="modalLoginForm"><input type="hidden" name="request_token"
value="aad-GRhc42RQ2jSQY1Vc6xkYgVNXZQsGi8hGLKNfp6Xd74fkx5Yup5lRNdwLl24GUdi2lw/MVRmS8vfOY6bPSVOjTLUHl6t7hACiyN+zOC5ENzuLZFVHXb+38t2/5HGoISCrJm98ZYzStnwj3VfH1tfP8ANGeIDk7V4xJyXtzoChldQS0OE+xG1g+uTIyAm5wVyfgiy9tMUO6kmvQkfbY3nJnz4EKsX75hJMkFVXvXEP4JUEmw08zn30j4n3UvJIRYX6bcxyAFcmwF35LaaHl9w="><input
type="hidden" name="urlAfterLogin" value="publication/320453789_A_generalized_simulation_development_approach_for_predicting_refugee_destinations"><input type="hidden" name="invalidPasswordCount" value="0"><input type="hidden" name="modalLogin"
value="yes">
<div class="nova-legacy-l-form-group nova-legacy-l-form-group--layout-stack nova-legacy-l-form-group--gutter-s">
<div class="nova-legacy-l-form-group__item nova-legacy-l-form-group__item--width-auto@m-up"><label
class="nova-legacy-e-text nova-legacy-e-text--size-m nova-legacy-e-text--family-sans-serif nova-legacy-e-text--spacing-xxs nova-legacy-e-text--color-inherit nova-legacy-e-label" for="input-modal-login-label"><span
class="nova-legacy-e-label__text">Email <div class="lite-page-tooltip "><span class="nova-legacy-e-text nova-legacy-e-text--size-m nova-legacy-e-text--family-sans-serif nova-legacy-e-text--spacing-none nova-legacy-e-text--color-grey-500">·
Hint</span>
<div class="lite-page-tooltip__content lite-page-tooltip__content--above">
<div class="nova-legacy-e-text nova-legacy-e-text--size-s nova-legacy-e-text--family-sans-serif nova-legacy-e-text--spacing-none nova-legacy-e-text--color-inherit"><b>Tip:</b> Most researchers use their institutional email address as
their ResearchGate login</div>
<div class="lite-page-tooltip__arrow lite-page-tooltip__arrow--above">
<div class="lite-page-tooltip__arrow-tip"></div>
</div>
</div>
</div></span></label><input type="email" required="" placeholder="Enter your email" id="input-modal-login" name="login" autocomplete="email" tabindex="1"
class="nova-legacy-e-input__field nova-legacy-e-input__field--size-m nova-legacy-e-input__ambient nova-legacy-e-input__ambient--theme-default"></div>
<div class="nova-legacy-l-form-group__item nova-legacy-l-form-group__item--width-auto@m-up">
<div class="lite-page-modal__forgot"><label class="nova-legacy-e-text nova-legacy-e-text--size-m nova-legacy-e-text--family-sans-serif nova-legacy-e-text--spacing-xxs nova-legacy-e-text--color-inherit nova-legacy-e-label"
for="input-modal-password-label"><span
class="nova-legacy-e-label__text">Password</span></label><a class="nova-legacy-e-link nova-legacy-e-link--color-blue nova-legacy-e-link--theme-bare lite-page-modal__forgot-link" href="application.LostPassword.html">Forgot password?</a>
</div><input type="password" required="" placeholder="" id="input-modal-password" name="password" autocomplete="current-password" tabindex="2"
class="nova-legacy-e-input__field nova-legacy-e-input__field--size-m nova-legacy-e-input__ambient nova-legacy-e-input__ambient--theme-default">
</div>
<div class="nova-legacy-l-form-group__item nova-legacy-l-form-group__item--width-auto@m-up"><label class="nova-legacy-e-checkbox"><input type="checkbox" class="nova-legacy-e-checkbox__input" aria-invalid="false" checked="" value="yes"
name="setLoginCookie" tabindex="3"><span class="nova-legacy-e-checkbox__checkmark"></span><span class="nova-legacy-e-checkbox__label"> Keep me logged in</span></label></div>
<div class="nova-legacy-l-form-group__item nova-legacy-l-form-group__item--width-auto@m-up"><button
class="nova-legacy-c-button nova-legacy-c-button--align-center nova-legacy-c-button--radius-m nova-legacy-c-button--size-m nova-legacy-c-button--color-blue nova-legacy-c-button--theme-solid nova-legacy-c-button--width-full" type="submit"
width="full" tabindex="4"><span class="nova-legacy-c-button__label">Log in</span></button></div>
<div class="nova-legacy-l-form-group__item nova-legacy-l-form-group__item--width-auto@m-up">
<div
class="nova-legacy-l-flex__item nova-legacy-l-flex nova-legacy-l-flex--gutter-m nova-legacy-l-flex--direction-column@s-up nova-legacy-l-flex--align-items-center@s-up nova-legacy-l-flex--justify-content-flex-start@s-up nova-legacy-l-flex--wrap-nowrap@s-up">
<div class="nova-legacy-l-flex__item">
<div class="nova-legacy-e-text nova-legacy-e-text--size-s nova-legacy-e-text--family-sans-serif nova-legacy-e-text--spacing-none nova-legacy-e-text--color-inherit">or</div>
</div>
<div class="nova-legacy-l-flex__item">
<div
class="nova-legacy-l-flex__item nova-legacy-l-flex nova-legacy-l-flex--gutter-m nova-legacy-l-flex--direction-column@s-up nova-legacy-l-flex--align-items-center@s-up nova-legacy-l-flex--justify-content-flex-start@s-up nova-legacy-l-flex--wrap-nowrap@s-up">
<div class="nova-legacy-l-flex__item">
<a href="connector/google"><div style="display:inline-block;width:247px;height:40px;text-align:left;border-radius:2px;white-space:nowrap;color:#444;background:#4285F4"><span style="margin:1px 0 0 1px;display:inline-block;vertical-align:middle;width:38px;height:38px;background:url('images/socialNetworks/logos-official-2019-05/google-logo.svg') transparent 50% no-repeat"></span><span style="color:#FFF;display:inline-block;vertical-align:middle;padding-left:15px;padding-right:42px;font-size:16px;font-family:Roboto, sans-serif">Continue with Google</span></div></a>
</div>
</div>
</div>
<div class="nova-legacy-l-flex__item">
<div class="nova-legacy-e-text nova-legacy-e-text--size-s nova-legacy-e-text--family-sans-serif nova-legacy-e-text--spacing-none nova-legacy-e-text--color-grey-500" align="center">No account?
<a class="nova-legacy-e-link nova-legacy-e-link--color-blue nova-legacy-e-link--theme-decorated" href="signup.SignUp.html?hdrsu=1&_sg%5B0%5D=a9m9mi3vE2t1P2FaM2fbwH5RWEv2UGfdcQ_BS7O3fCO3E8jfhb04jijuCW0qBHu-pFlNBOoeOZO167r4CdRE2jPplQc">Sign up</a>
</div>
</div>
</div>
</div>
</div>
</form>Text Content
YOUR PRIVACY CHOICES We and our partners store and access non-sensitive information from your device, like cookies, and process personal data, like IP addresses and unique identifiers to personalize content and ads, measure performance, and analyze audiences. By clicking Accept, you consent to this data collection and processing by us and our 200 partners. You can select Reject to continue with only strictly necessary cookies or Customize to manage your preferences. Some processing of your personal data may not require your consent, but you have a right to object to such processing. You can withdraw your consent at any time from the consent preferences link in the footer of any ResearchGate page. For more information, see our Privacy Policy. We and our partners process data for the following purposesPersonalised advertising and content, advertising and content measurement, audience research and services development , Precise geolocation data, and identification through device scanning, Store and/or access information on a device CustomizeRejectAccept ArticlePDF Available A GENERALIZED SIMULATION DEVELOPMENT APPROACH FOR PREDICTING REFUGEE DESTINATIONS Scientific Reports * October 2017 * 7(1) DOI:10.1038/s41598-017-13828-9 * License * CC BY 4.0 Authors: Diana Suleimenova * Brunel University London David Bell * Brunel University London Derek Groen * Brunel University London Download full-text PDFRead full-text Download full-text PDF Read full-text Download citation Copy link Link copied -------------------------------------------------------------------------------- Read full-text Download citation Copy link Link copied Citations (90) References (53) Figures (6) ABSTRACT AND FIGURES In recent years, global forced displacement has reached record levels, with 22.5 million refugees worldwide. Forecasting refugee movements is important, as accurate predictions can help save refugee lives by allowing governments and NGOs to conduct a better informed allocation of humanitarian resources. Here, we propose a generalized simulation development approach to predict the destinations of refugee movements in conflict regions. In this approach, we synthesize data from UNHCR, ACLED and Bing Maps to construct agent-based simulations of refugee movements. We apply our approach to develop, run and validate refugee movement simulations set in three major African conflicts, estimating the distribution of incoming refugees across destination camps, given the expected total number of refugees in the conflict. Our simulations consistently predict more than 75% of the refugee destinations correctly after the first 12 days, and consistently outperform alternative naive forecasting techniques. Using our approach, we are also able to reproduce key trends in refugee arrival rates found in the UNHCR data. Overview of geographic network models for (a) Burundi, (b) Central African Republic and (c) Mali. Models contain conflict zones (red circles), refugee camps (dark green circles), forwarding hubs (light green circles) and other major settlements (yellow circles). Interconnecting roads are given in a simplified straight-line representation, with adjacent blue numbers used to indicate their length in kilometres. Background maps are courtesy of https://carto.comcarto.com created using OpenStreetMap data that is further modified with the use of https://inkscape.org/en/release/0.91Inkscape0.91. … Comparison of number of refugees in camps between the simulation and the data (left column), and overview of the averaged relative difference between simulation and data (right column). The averaged relative difference across camps between simulation and data is given by the red line. We provide these comparisons respectively for (a,b) the Burundi simulations (top row), (c,d) the CAR simulations (middle row) and (e,f) the Mali simulations (bottom row). … Number of refugees as predicted by our simulation and obtained from the UNHCR data for the Burundi conflict. (a-e) Graphs are ordered by camp population size, with the most populous camp on the top to the smallest one on the bottom. … Number of refugees as predicted by our simulation and obtained from the UNHCR data for the CAR conflict. (a-h) Graphs are ordered by camp population size, with the most populous camp on the top to the smallest one on the bottom (see remaining six camps in Fig. S1). … +1 Number of refugees as predicted by our simulation and obtained from the UNHCR data for the Mali conflict. (a-g) Graphs are ordered by camp population size, with the most populous camp on the top to the smallest one on the bottom. … Figures - available from: Scientific Reports This content is subject to copyright. Terms and conditions apply. Discover the world's research * 25+ million members * 160+ million publication pages * 2.3+ billion citations Join for free Publisher Full-text 1 Public Full-text 1 Access to this full-text is provided by Springer Nature. Learn more Content available from Scientific Reports This content is subject to copyright. Terms and conditions apply. 1 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 www.nature.com/scientificreports A generalized simulation development approach for predicting refugee destinations Diana Suleimenova1, David Bell1 & Derek Groen1,2 In recent years, global forced displacement has reached record levels, with 22.5 million refugees worldwide. Forecasting refugee movements is important, as accurate predictions can help save refugee lives by allowing governments and NGOs to conduct a better informed allocation of humanitarian resources. Here, we propose a generalized simulation development approach to predict the destinations of refugee movements in conict regions. In this approach, we synthesize data from UNHCR, ACLED and Bing Maps to construct agent-based simulations of refugee movements. We apply our approach to develop, run and validate refugee movement simulations set in three major African conicts, estimating the distribution of incoming refugees across destination camps, given the expected total number of refugees in the conict. Our simulations consistently predict more than 75% of the refugee destinations correctly after the rst 12 days, and consistently outperform alternative naive forecasting techniques. Using our approach, we are also able to reproduce key trends in refugee arrival rates found in the UNHCR data. Global forced displacement has reached record levels. In 2017, 65.6 million people were forcibly displaced world- wide, a number which includes 22.5 million refugees1. Common causes of forced migration include push and pull characteristics, such as the present social, political, and economic conditions of migrants’ origin and potential destination, as well as intervening characteristics between these two locations2,3. Migration is a complex phenom- enon and the push-pull characteristics can be insucient to explain forced migration4. Several groups identied sets of other causal factors that lead to forced displacement, including conicts, ethnic or religious dierences, and existential obstacles such as severe ecological decline5,6. Previous studies have shown that the inuence of these causal factors can be determined using migration ow models. For instance, Shellman and Stewart7 investigated Haitian migration to the United States using an early warning model of forced migration and predicted risk factors, such as civil violence, economic conditions and external interventions, that forced people to migrate. Similarly, Martineau8 used an early warning model to predict which countries have the potential to create refugees. However, existing early warning models of forced migration focus on understanding the causes9 and are not as successful in predicting refugee movements as in predicting natural disasters10–12. Moreover, they lack the accuracy and exibility to accommodate the context changes that lead to large-scale refugee movements13. As a result, there are relatively few appropriate models for predicting refugee movements14,15. Computational models have been widely applied to study migration processes16. Moreover, they have the potential to contribute to a better understanding of refugee movement patterns, and to inform, predict and full gaps within forced migration estimations17. In particular, computational models could be applied interactively to assist governments and organisations in estimating where and when refugees are likely to arrive18, and which camps are most likely to become full in the short term. Simulating refugee movements also has potential due to its reduced ethical burden, which normally impedes empirical analysis, and the possibility to derive causal relations17. Agent-based modelling (ABM) is a popular simulation approach that can explicitly model social interactions and networks emerging from it. Hence, ABM is becoming a prominent method for population and migration studies (e.g.,17,19–21). In addition, it is popular due to its decentralized approach22, which allows a heterogeneous mix of many agents to act and interact autonomously, leading to emergent behaviours in the system at higher 1Brunel University London, Department of Computer Science, London, UB8 3PH, United Kingdom. 2University College London, Centre for Computational Science, London, WC1H 0AJ, United Kingdom. Correspondence and requests for materials should be addressed to D.G. (email: derek.groen@brunel.ac.uk) Received: 4 July 2017 Accepted: 27 September 2017 Published: xx xx xxxx OPEN Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 2 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 levels22–24. ABM is especially suitable for modelling active objects, such as individuals, animals or products, in relation to time, events or behaviour25, and it has been applied to model problems ranging from small-scale behavioural dynamics to large scale migration simulations26,27. However, a few important challenges have been identied within the ABM community. For instance, there is an ongoing debate on whether prediction should be a major purpose for ABMs28, or whether explaining and illu- minating problems should be a priority29. Specically for migration studies, Klabunde and Willekens22 identify major challenges in both the denition of decision-making theories and the selection of empirical evidence for model validation. ABMs are already used in a wide range of refugee-related settings, such as disaster-driven migration which incorporate changes in climate and demographics30. For example, Hassani-Mahmooei and Parris31 analysed the inuence of climate change on migration in Bangladesh while Kniveton et al.19,32 developed an ABM to sim- ulate climate migration in Burkina Faso between 1970–2000 and to predict future migration ows to 2060. Additionally, Anderson et al.33,34 suggested an ABM for refugee communities to inform policy decisions for gov- ernments and other organisations. e German armed forces developed an ABM to understand interactions and behaviour of refugees with military groups in a refugee camp environments20. In the context of predicting and forecasting refugee movements, Sokolowski and Banks35 developed an ABM Environment Matrix that can be used to accurately represent irregular migration movements using simulation. Similarly, Latek et al.36 build a multi-agent model that predicts the Syrian conict characteristics and investigated potential conditions and outcomes of the conict. Hattle et al.37 examined the Syrian refugee ows to European countries using ABM and discussed possible policy recommendations on distributing humanitarian resources amongst potential refugee hosting countries. Several groups also applied ABM to capture local aspects, such as networks, group formation and travel distance in the refugee crisis and stress the importance of computational modelling for migration predictions21,38,39. In this work, we present a generalized simulation development approach (SDA) to predict the distribution of refugee arrivals across camps, given a particular conict situation. Our SDA has six phases, and is partially based on the notion of the Simplied Simulation Development Process, presented by Heath et al.40. It encompasses the formulation of the problem (phase 1), the translation into a computer model (phase 2,3 and 4) and the opera- tional validation (phase 5 and 6). e conceptual validation does not pertain a specic phase, as we present a conceptual model that can be readily adopted as part of the SDA. Our main reason to develop a full SDA, in contrast with merely proposing a simulation model design, is the need for organizations to facilitate rapid simulation development when a conict occurs. In conict situations, a model design alone would contain too little information to facilitate rapid development, as such development activities inevitably involve the selection of data sources, the extraction and conversion of data, as well as the val- idation of simulation predictions against empirical data. We provide a diagrammatic overview of our SDA, and its six phases, in Fig.1. Here, in the rst phase, we select a country and time period of a specic conict which resulted in large scale forced migration. In the second phase, we obtain relevant data to the conict from three data sources: the Armed Conict Location and Events Database (ACLED, http://www.acleddata.com/data/acled-version-7-1997-2016/ acleddata.com)41, the UNHCR database (http://data2.unhcr.org/en/situationsdata2.unhcr.org), and the Bing Maps platform (https://www.bing.com/mapsbing.com/maps). We use ACLED to obtain the locations and dates of battles that have taken place in the conict, and the UNHCR database to obtain the number of refugees in the conict, as well as the camp locations and capacities. We rely on the Bing Maps platform to obtain locations of major settlements and routing information between the various camps, conict zones and other settlements. We provide a detailed description of the data collection procedure in the Methods section. In the third phase, we construct our initial simulation model using these data sets, and create among other things a network-based ABM model. We present the three network-based ABM models, one for each conict we seek to model, in Fig.2, while we present our detailed assumptions in the Methods section and provide our source data as part of Supplementary Note 2. Once we have constructed the initial mode, we rene it as part of Figure 1. Simulation development approach for predicting the distribution of refugee arrivals across camps. Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 3 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 the fourth phase. Here, we manually extract population data to help determine where refugees ee from (see Methods section for details), as well as information on border closures and forced redirections of refugees (see Supplementary Note2). e h phase involves the main simulation, which we run to predict, given a total number of refugees in the conict, the distribution of refugees across the individual camps. We run our simulations using the FLEE simu- lation code. FLEE is optimised for simplicity and exibility and provides a range of scripts to handle and convert refugee data from the UNHCR database. As part of this work, we publicly release the FLEE code, as well as all our input and output data sets, under a BSD 3-clause license (see the subsections on Code Availability and Data Availability in the Methods section). Once the simulations have completed, we analyse and validate the results against the full UNHCR refugee numbers as part of the sixth phase (see Results section for several examples). To showcase the added value, and generalized nature, of our SDA, we apply it to model three refugee crises in African countries. ese crises include the 2015–2016 civil war in Burundi and the 2013–2016 conict in the Central African Republic (CAR), both which to our knowledge have never been modelled before. We also model the Northern Mali conict in 2012–201342,43, which we have previously modelled in rudimentary form (see Supplementary Note5). ese three African countries demonstrate dierent conict initiation scenarios, but they all have common drivers forcing people to ee, such as political instabilities, violence and civil war. According to Turchin44, when countries experience long-term pressures they result in civil wars, social and political instabilities. By understand- ing historical data of these socio-political instabilities and political violence events, it is possible to nd patterns that explain their cause and time of occurrence45–47. ough a full historical analysis of the three conicts of interest is well beyond the scope of this work, we do provide a brief summary of each conict in Supplementary Note1. e Burundian crisis was triggered by the third-term election of President Pierre Nkurunziza in April Figure 2. Overview of geographic network models for (a) Burundi, (b) Central African Republic and (c) Mali. Models contain conict zones (red circles), refugee camps (dark green circles), forwarding hubs (light green circles) and other major settlements (yellow circles). Interconnecting roads are given in a simplied straight-line representation, with adjacent blue numbers used to indicate their length in kilometres. Background maps are courtesy of https://carto.comcarto.com created using OpenStreetMap data that is further modied with the use of https://inkscape.org/en/release/0.91Inkscape0.91. Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 4 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 2015. His election triggered protests, coups and eventually a refugee crisis48,49. We choose to simulate this conict from the start of the refugee crisis, around the 1st of May 2015, until the 31st of May 2016, for a period of 396 days. In CAR, the Seleka group (Muslim population) overthrew the central government, in March 201350. Not long aer, anti-Balaka (Christian militia groups) took over the power. Muslim and Christian communities started a long string of conicts and violent attacks51. e crisis continued for several years and to capture it simulation period is 820 days from 1 December 2013 to 29 February 2016. In the case of Mali, the crisis was due to insur- gent groups, who began a campaign to ght for the independence of the Azawad region. e conict started on the 16th January 2012, when Touareg rebels began conquering settlements in Northern Mali17. In this case, we selected a simulation period of 300 days, from the 29th of February 2012, when the rst camp registrations were recorded, until 25th December 2012, when the vast majority of refugees had been registered in the camps. Results We present results from our SDA, which we applied to predict the distribution of refugees across camps in three African conicts. For each conict, we compare our prediction results with the UNHCR refugee camp registra- tion data. We provide a list of the refugee camps in each conict in Table1. We also present several error measures in Fig.3, including an overview of the number of refugees in camps according to the simulation and the UNHCR data in (Fig.3a,c and e) and the averaged relative dierence between the simulation results and the UNHCR data (explained in the Methods Section) in Fig.3b,d and f. e averaged relative dierence is less than 0.5 aer the rst few days, indicating that our simulations accurately predict more than 75% of the refugee movements in absolute terms. In all our runs, the averaged relative dierence is lower at later stages of the simulations, with relative dierences of 0.1–0.3 or towards the end of all runs. Burundi. We present our simulation predictions and the UNHCR refugee counts for the Burundi conict in Fig.4. Within the camps in Nyarugusu, Mahama and Nakivale, our simulation results accurately capture the key growth trends in refugees. Our approach does underpredict the refugee population growth in Mahama, as there is a delay in refugee arrival due to the many non-conict settlements between Mahama and the conict zones. Both the Nduta and Lusenda camps opened only aer the start of the period of simulation. Nduta was only established as a refugee camp on the 10th of August 2015 (day 101), aer Nyarugusu became overpopulated. In the case of Nduta, our simulation shows a small population of travelling refugees at the start (when the location was not yet a camp), and a steep population increase to 30,000 during the 90 days aer the camp is opened. Lusenda, which opened on day 90, quickly lls to capacity in the simulation, whereas a more gradual increase can be observed in the data. Here, the mismatch could be due to delays in the UNHCR registration process, as virtually no refugees were properly registered in the whole of DRC prior to the 30th of October 2015 (day 182). For Burundi (Fig.3a), our simulations contain substantially fewer refugees in camps than the UNHCR meas- urements for the same day. is dierence is larger than in other cases and aects the averaged relative dierence (Fig.3b), primarily because Burundi is a densely populated country with a large number of settlements in the net- work graph. However, the dierence decreases aer Day 5 once substantial numbers of refugees arrive in camps in the simulation, and only increases to a peak around 0.48 on Day 151, due to a coincidence of peak mismatches in both Nyarugusu and Nduta. Central African Republic. In Fig.5 we present the number of refugees in camps for the CAR conict sim- ulation. Our simulation predictions closely follow the trends observed in the data for the two largest camps, East Congo and Adamaoua. Here our simulation underpredicts the total refugee population in East Congo by about 35,000 (~20%), and overpredicts the population in Adamaoua by about 23,000 (~30%). e camps in DRC (Inke, Mole, Boyabu and Mboti) were subject to border closures between CAR and DRC from the 5th of December 2013 (simulation day 4), until the 30th of June 2014 (day 211, see Supplementary Note2 for details). is is reected by a period of relatively stable refugee populations both in the simulation and in the data. Bili also is located within DRC, but was established only aer the border was reopened. Country Neighbouring country Camps Burundi Tanzania Nyarugusu and Nduta Rwanda Mahama Uganda Nakivale Democratic Republic of the Congo (DRC) Lusenda CAR Cameroon East and Adamaoua Chad Belom, Dosseye, Amboko, Gondje and Moyo DRC Inke, Mole, Bili, Mboti and Boyabu Republic of the Congo (RC) Betou and Brazaville Mali Mauritania Mbera Burkina Faso Mentao and Bobo-Dioulasso Niger Abala, Mangaize, Niamey and Tabareybarey Table 1. List of existing camps used in simulations. Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 5 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 e predicted refugee counts in the Chad camps (Amboko, Belom, Dosseye and Gondje) are in close agree- ment with the data, except that large uctuations occur during the simulation aer the border closure on the 12th of May 2014 (day 163). At this time all the camps are close to full occupancy, which results in refugees moving from between the camps and the city of Gore, a city in Chad which lies in close proximity to the camps. e Betou camp in Congo is an another example of a camp close to the conict areas, and it also lls up quickly in the simulation. e Brazaville location is far removed from the conict zone, and here our simulation underpredicts the refugee population. It could be that the size of the city of Brazaville may increase its attractive- ness as a refugee destination. We did not incorporate this factor in the runs presented here, but we do wish to examine it in future simulation studies. In the CAR situation (Fig.3c and d) the mismatch in the number of refugees remains relatively small, while the averaged relative dierence uctuates around 0.3. e jump in error around Day 300 is largely due to a sudden large increase in refugees in East Cameroon at that time, according to the UNHCR data. Mali. In Fig.6 we present the number of refugees in camps around Mali over the 300 day simulation period. Our simulation results are in close agreement with the data for the two largest camps. e maximum dierences here are an underprediction of 7,000 (~18%) for Mbera around day 135, and an overprediction of about 4,500 (~60%) for Abala around day 160. Tabareybarey, Niamey, Mentao and Bobo-Dioulasso were established once the conict was already underway. Tabareybarey and Niamey camps have refugees for simulation and data from day 30, whereas the camps in Burkina Faso, Mentao and Bobo-Dioulasso, reopened their previously closed borders on the 1st of April 2012 (day 32). Figure 3. Comparison of number of refugees in camps between the simulation and the data (le column), and overview of the averaged relative dierence between simulation and data (right column). e averaged relative dierence across camps between simulation and data is given by the red line. We provide these comparisons respectively for (a,b) the Burundi simulations (top row), (c,d) the CAR simulations (middle row) and (e,f) the Mali simulations (bottom row). Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 6 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 e simulation predicts a fast-paced growth of refugee population for both Mentao and Bobo-Dioulasso, while the data features a sudden spike in refugee arrivals around day 30 in these camps. e simulation pre- dictions for Mangaize in Niger are in line with the data, though slightly higher. e large inow early in the simulation is primarily due to the close proximity of Mangaize to one of the early conict zones (Menaka). Our simulation results do not accurately match the data for Tabareybarey and Niamey. Niamey is not directly con- nected to regions in Mali, due to two other refugee camps being located along the way. However, Niamey is a large capital city (like Brazaville in the CAR simulation) which may be the reason why more refugees choose that des- tination than our simulation predicts. In general, our predictions overestimate the refugee inow into the three border camps in Niger. An important cause here may be the presence of partial restrictions for crossing the Niger border during the conict17. In the Mali situation (Fig.3e and f) we see a large but decreasing mismatch at the very start of the simulation. is is because the Fassala camp is technically not dened as a camp within our simulation, as refugees were already redirected from Fassala to Mbera from the start of the simulation period. However, Fassala is considered to be a camp according to the data. Aer Day 30, the number of refugees in camps in the simulation is relatively close to the reported number, and the averaged relative dierence remains relatively constant. Comparison with naive prediction models. To our knowledge, there are no other prediction techniques that have been previously applied in this setting. However, it is possible to perform naive predictions, extrapolat- ing future behaviour from historical data, aer a conict has started. To measure the added value of our prediction approach, we here present a comparison of our method against a set of naive prediction models. We compare the accuracy of our method by obtaining the Mean Absolute Scaled Error (MASE) relative to the six other techniques (see methods section for denition). e MASE was rst pro- posed by Hyndman et al.52, and is well suited to quantify simulation accuracy due to its scale invariant nature and Figure 4. Number of refugees as predicted by our simulation and obtained from the UNHCR data for the Burundi conict. (a–e) Graphs are ordered by camp population size, with the most populous camp on the top to the smallest one on the bottom. Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 7 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 the fact that it symmetrically penalizes overestimations and underestimations. In addition, the MASE is straight- forward to interpret: in our case its value is less than one if our prediction approach has a smaller error, while its value is more than one if the selected naive technique as a smaller error. For comparison purposes, we have created three dierent types of naive models, all of which rely on some section of historical data to extrapolate values in the future. While our simulation approach can be used from Day 1 to provide a prediction of camp refugee populations, we can only apply naive models aer a number of days have elapsed. is is because naive models extrapolate from past data; and such data can only be acquired aer the conict has started and refugees have departed. Figure 5. Number of refugees as predicted by our simulation and obtained from the UNHCR data for the CAR conict. (a–h) Graphs are ordered by camp population size, with the most populous camp on the top to the smallest one on the bottom (see remaining six camps in Fig.S1). Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 8 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 In this section we compare our approach, as described in the main paper, against naive model predictions that take place respectively 7 days, and 30 days aer the starting date of the respective simulation periods. We argue that a week is required to obtain sucient data to apply any kind of meaningful extrapolation. However, naive models that require more than a month before they can be applied are arguably of little use, as many of the initial refugee movements have already taken place by then (particularly in the case of the Burundi conict). It should be noted that the collection of refugee registrations is by no means an instantaneous process, and any time overhead in obtaining such would further delay the application of these naive models. Figure 6. Number of refugees as predicted by our simulation and obtained from the UNHCR data for the Mali conict. (a–g) Graphs are ordered by camp population size, with the most populous camp on the top to the smallest one on the bottom. Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 9 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 For each refugee camp location in each conict, we have applied the following three types of naive prediction: • 0th order (at) extrapolation: Here we take the refugee count on either day 7 or day 30 in each camp, and assume that this number does not change over time. • 1st order (sloped) extrapolation: Here we take the refugee count on either day 7 or day 30, as well as the reg- istration count on day 0. We then linearly extrapolate future values in time from these two registration counts. • Extrapolation by ratio (fraction): Here we take the refugee fraction in a given camp, which we calculate by dividing the refugee count in a given camp, on either day 7 or day 30, by the total number of refugees across all camps on that same day. We then forecast refugee counts in each camp by assuming that this refugee fraction remains constant over time, and predict future value by taking that xed fraction of the total refugee popula- tion (which is a known quantity in our setting) over time. We present the results from our comparison in Table2. In all cases, our prediction approach results in a lower averaged relative dierence than the naive prediction models. We obtained MASE scores of 0.0639-0.942 (Burundi), 0.0367-0.705 (Central African Republic), and 0.116-0.513 (Mali). Discussion We have presented a generalized simulation development approach (SDA) for predicting the distribution of incoming refugees across destination camps. Accurate predictions can help save refugee lives, as it helps govern- ments and NGOs to correctly allocate humanitarian resources to refugee camps before the (oen malnourished or injured) refugees themselves have arrived. To our knowledge, we are the rst to attempt such predictions across multiple major conicts using a single simulation approach. Using our approach, we have reproduced the key refugee movement patterns in each of the three conicts and correctly predicted at least 75% of the refugee movement destinations in all these conicts aer the rst 12 days. In the Burundi conict, our approach correctly predicts the largest inows in Nyarugusu, Mahama and Nakivale during the early stages of the conict. In CAR, our prediction approach correctly reproduces the growth pattern in East Congo, as well as the stagnation of refugee inux in the Chad camps. In the case of Mali, our predictions accurately capture the trends in the data for both Mbera and Abala, which together already account for ~75% of the refugee population. Our results are insensitive to most simulation parameter changes, with the notable exception that increasing the probability for refugee agents in non-conict/non-camp locations actually results in a further reduced error (see Methods section for a summary and Supplementary Note 4 for a detailed discussion regarding this). As a result of conducting this study, we discovered several important issues and limitations. For example, our model omits a range of factors which are considered important according to the empirical literature, but for which we could not nd accurate and tractable means to convert empirical conclusions to simulation parameters. In some cases such as GDP and presence of existing conicts, the signicance of these factors has been conrmed on a country-by-country level but not on a city-by-city level3,53. In other cases, such as religion and ethnicity, we simply did not nd reliable statistical information on a local level for these conicts. Some parameters, such as the level of knowledge of refugee agents about the surrounding region, were found to have little eect on the simula- tion results beyond being aware of adjacent locations (see TableS5). e obtained averaged relative dierence also changes little when we adjust maximum movement speed of refugees to values less or more than 200 kilometres per day (see TableS4). In general, empirical data collection during these conicts is very challenging, in part due to the nature of the environment and in part due to the severe and structural funding shortages of UNHCR emergency response missions. Both CAR and Burundi are among the most underfunded UNHCR refugee response operations, with funding shortages of respectively 76 and 62%54. More funding for these operations are bound to save human lives, have the side benet of providing more comprehensive empirical data, and thereby enable the validation of more detailed prediction models. An additional important element that is absent in all our data sources is indications of the level of data-related uncertainty. Knowledge of this level of uncertainty would allow us to accurately quantify how uncertainty from Run name MASE 7 Day MASE 30 Day at slope fraction at slope fraction Burundi 0.279 0.144 0.942 0.443 0.0639 0.791 CAR 0.585 0.705 0.452 0.639 0.0367 0.341 Mali 0.23 0.127 0.503 0.314 0.116 0.513 Weighted average 0.453 0.473 0.598 0.542 0.0544 0.491 Table 2. Comparison of our prediction approach against six naive models for each of our three conict simulations. Simulations were run with the same settings as those presented in the main paper. We report on the Mean Averaged Scaled Error for each of the naive models in columns 2 to 7. Here values below 1 indicate that forecasts using our prediction approach resulted in a smaller averaged relative dierence than those relying on that specic naive model. In the bottom row we provide a weighted average of the MASE score across the three conict simulations, with the weightings based on the maximum number of refugees in each conict (205445 for Burundi, 424496 for CAR, and 89991 for Mali). Please refer to the Methods Section for details on the six naive models. Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 10 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 the source data aects the overall outcome of our simulations, the quality of our validation tests, and the perfor- mance of our approach versus that of naive models. Yet, important steps have been made in recent years, as the combination of a conicts database41, a public UNHCR refugee data repository and a sophisticated mapping platform enabled us to do this work. And given the increasing eort in collecting refugee data, and increasing recognition for data science, we are condent that future research eorts on modelling refugee movements will be accelerated by ongoing advances in data collection. Methods Processing input and validation refugee data. To obtain our input data, we took the following steps. First, we selected three conicts that featured on data2.unhcr.org (accessed June 2016) and manually obtained the refugee registration data for each camp from the website in comma separated value (CSV) format. We rened the data by interpolating linearly between data points and calculated the total refugee count by aggregating the (interpolated) registrations for each of the camps. e source data includes level 1 refugee registrations and, aer certain dates, level 2 registrations. As level 1 registrations are known to result in overestimations of refugee count, we scaled down these values such that the last data point using level 1 refugee registrations matches the rst data point using level 2 registrations. We exclude Internally Displaced People from the model, as there is a lack of sys- tematic data providing their exact destinations in our scenarios. We obtain conict locations from the ACLED database, omitting settlements with less than 10,000 inhabit- ants, and noted the start date of any event labelled as “battle” during the simulation period. Locations are labelled as a conict zone as soon as such an event has occurred. All conict locations are assigned a population based on the latest census data. Constructing the network graphs. We provide detailed network graphs in Fig.2. We selected locations by combining our ACLED conict locations and UNHCR camp locations with major settlements that reside en-route between these locations. Locations are interconnected with links in cases where we noticed the presence of roads in Bing Maps, the length of the link (in km) was then estimated using the Bing route planner for cars. In cases where obvious shorter routes were visible, we dragged the Bing marker to force the soware to calculate this shorter route. To retain the simplicity of our model, and to reect the frequent occurrence of direct redirections of refugees to camps, we directly connected refugee camps to the nearest location in the country of conict. In some cases, we added “forwarding” locations, where refugees are automatically rerouted to a camp, or opened camps aer the start of the simulation, following descriptions in UNHCR reports (see Supplementary Note2 for details). We also removed links when border closures were reported by the UNHCR, and added a link aer the start of the simulation when a border opening was reported (see Supplementary Note2 for details). Choosing simulation parameters and assumptions. We provide a owchart of the key elements in our simulation algorithm in Supplementary Note6. Each step of the simulation represents one day. During each step, we insert a number of refugees into the simulation based on the daily increase in the total refugee regis- tration count from the UNHCR data. ese refugees are inserted in their location of origin, which is one of the conict locations (as obtained from the ACLED database, see section “Processing input and validation refugee data”). e exact location is picked among all conict zones, where the likelihood of each conict zone being selected is proportional to its population. e population of a location is decremented by one each time a refugee agent is created. Because we insert refugees in conict zones on the day of camp registration and refugee travel is non-instantaneous, our simulation approach normally results in an under-prediction of the number of refugees. To correct for this, we multiply the refugee populations in each of the camps by Ndata,all/Nsim,all, where Ndata,all is the total refugee count for the conict on a given day according to the UNHCR data. In our setting, this is a known quantity, as we are predicting the distribution of refugees across camps, given this total refugee count. Nsim, all is the total number of refugees in camps according to the simulation on that same day. We discuss and measure the eect of using this correction in Supplementary Note3. We did not rescale our output when comparing the number of agents in the simulation and the data (see Fig.3a,c and e). Decreases in UNHCR refugee registrations increment a “refugee debt” variable, which rst needs to be compensated by subsequent registration increases before additional agents are again inserted in the simulation (i.e., we do not delete agents). During each step, a refugee agent can traverse zero, one or more links. e probability of traversing a link is determined by the move chance, which we initially set at 1.0 for refugees in transit between locations, 1.0 for refu- gees in conict locations, 0.001 for those in refugee camps, and 0.3 for those in all other locations. As we could not nd empirical evidence supporting these parameters, we initially chose these parameters based on our intuitions, and performed our main simulations using these initial choices (i.e., we did not optimize our parameter choices to minimize the error, as we believe such parameter tting could reduce the applicability of our approach to other conict situations). Aer the main run was performed, we analyzed the sensitivity of each of these parameters (see Supplementary Note4 for details). To summarize this analysis, we found that our results are insensitive to the conict location move chance parameter across the full tested range, and insensitive to the refugee camp move chance for values ≤0.01 (which implies the assumption that refugees remain in a refugee camp, on average, for 100 days or more). However, our simulations results did show sensitivity to the move chance for all other loca- tions, with higher move chances resulting in smaller validation errors, and lower move chances resulting in larger validation errors. We reect on the implications of this parameter sensitivity in detail in Supplementary Note4. When an agent traverses a link (with the probability determined by the aforementioned move chance) it needs to choose one of the available paths. Path selection is done using a weighted probability function, the weight of each link being equal to the attractiveness value of the destination divided by the length of the link in kilometres. e attractiveness value of the destinations equals 0.25 for conict zones, 1.0 for other locations in the country of Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 11 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 conict, and 2.0 for locations abroad. Again, these values were initially chosen based on our own intuition, with the sensitivity being analyzed aer the main run was performed (see Supplementary Note4). In the case of these two parameters (attractiveness value for camps, and for conict zones), we found that none of these parameters had a signication eect on the accuracy of our simulation. We also assumed that refugees travel no more than 200 km/day, and likewise found our simulation has low sensitivity to higher travel limits (see Supplementary Note4), though our error increases if we choose much lower travel limits. If a refugee reaches the end of a link but has travelled less than 200 km on that day, then a new move chance calculation (and possible move) is performed. In traversing between locations, refugees take major roads, which are shortest journey paths identied using route planners from https://www.bing.com/mapsbing.com/maps and https://www.google.co.uk/mapsgoogle. co.uk/maps Processing simulation output data. We calculate an averaged relative dierence using the following equation: =∑ |− | ∈ Et nn N () () (1) xS simxtdataxt data all ,, ,, , us, the number of refugees found in each camp x of the set of all camps S at time t is given by Nsim,x,t based on the simulation predictions, and by ndata,x,t based on the UNHCR data. e total number of refugees reported in the UNHCR data is given by Ndata,all. We also present comparisons to naive models using the Mean Absolute Scaled Error (MASE). We calculate the MASE score using the aforementioned averaged relative dierence at each time step, as follows: =∑ ∑ = − = MASE T Et Et 1 () () (2) t T Tw tw Tnaive 0 1 Here, T is the full duration of the simulation, w is the warmup period required for the naive model to make its predictions (in our case either 7 or 30 days, depending on the model type). e averaged relative dierence using the naive model compared to the validation data is given by Enaive(t). Code Availability. We use the FLEE17 for our simulations, which is an agent-based modelling code written in Python with a limited feature set that is optimised for simplicity and exibility. It is able to support simula- tions with 100,000 s of agents on a single desktop, and provides users with the ability to dene and use their own models through a relatively straightforward API. We provide a range of functional tests to allow users to verify the consistency of the code results. FLEE also features a range of scripts to handle and convert refugee data from data2.unhcr.org, as well as an automated plotting tool for output generated by the simulation. To use the code, one requires a Python 3 interpreter, as well as the numpy, scipy and pandas Python modules. As part of this publica- tion, we provide the version of FLEE we used to run these simulations. is version can be found at http://www. github.com/djgroen/ee-release, and is distributed under a BSD 3-clause license. Data Availability. All input and output data publicly available on Figshare with DOI https://doi. org/10.17633/rd.brunel.5446813.v1, under a CC-By 4.0 license. References 1. UNHC. Figures at a glance. United Nations High Commissioner for efugees. Available at: http://www.unhcr.org/u/gures-at-a- glance.html (2017). 2. Lee, E. S. A theor y of migration. Demography 3, 471–486 (1966). 3. Moore, W. H. & Shellman, S. M. Whither will they go? A global study of refugees’ destinations, 1965–1995. International Studies Quarterly 51, 811–834 (2007). 4. EASO. e push and pull factors of asylum-related migration. A literature review. European Asylum Support Oce (2016). 5. Wood, W. B. Forced migration: Local conicts and international dilemmas. Annals of the Association of American geographers 84, 607–634 (1994). 6. alipeni, E. & Oppong, J. e refugee crisis in Africa and implications for health and disease: A political ecology approach. Social Science & Medicine 46, 1637–1653 (1998). 7. Shellman, S. M. & Stewart, B. M. Predicting ris factors associated with forced migration: An early warning model of Haitian ight. Ci vi l Wa rs 9, 174–199 (2007). 8. Martineau, J. S. ed ags: A model for the early warning of refugee outows. Journal of Immigrant & Refugee Studies 8, 135–157 (2010). 9. Bunoiu, M.-D. & Udroiu, I. Spotting trouble in migration ows: An indicator-based early warning model. In Iancu, N. & Chiru, I. (eds) Proceedings of the XXIst International Conference. Intelligence in the Knowledge Society, 95–104 (Bucharest, 2016). 10. Schmeidl, S. & Jenins, J. C. The early warning of humanitarian disasters: Problems in building an early warning system. International Migration Review 32, 471–486 (1998). 11. Schmeidl, S. e early warning of forced migration: State or human security? In Refugees and Forced Displacement: International Security, Human Vulnerability, and the State, 130–155 (United Nations University Press, Toyo, 2003), newman, e. and selm j.v. (ed.) edn. 12. Birmann, J., Chang Seng, D. & Setiadi, N. Enhancing early warning in the light of migration and environmental shocs. Environmental Science and Policy 27S, S76–S88, https://doi.org/10.1016/j.envsci.2012.04.002 (2013). 13. Lopez-Lucia, E. Early warning models for irregular migration. Helpdes esearch eport. GSDC Applied nowledge Services (2015). 14. Edwards, S. Computational tools in predicting and assessing forced migration. Journal of Refugee Studies 21, 347–359 (2008). Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 12 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 15. Disney, G., Wiśniowsi, A., Forster, J. J., Smith, P. W. F. & Bija, J. Evaluation of existing migration forecasting methods and models. eport for the Migration Advisory Committee: Commissioned research. ESC Centre for Population Change, University of Southampton (2015). 16. Willeens, F. Migration ows: Measurement, analysis and modeling, 225–241, https://doi.org/10.1007/978-94-017-7282-2_11 (Springer Netherlands, Dordrecht, 2016). 17. Groen, D. Simulating refugee movements: Where would you go? Procedia Computer Science 80, 2251–2255 (2016). 18. Perez Estrada, L. E., Groen, D. & amirez-Marquez, J. E. A serious video game to support decision maing on refugee aid deployment policy. Procedia Computer Science 108, 205–214, https://doi.org/10.1016/j.procs.2017.05.112 International Conference on Computational Science, ICCS 2017, 12–14 June 2017, Zurich, Switzerland (2017). 19. niveton, D., Smith, C. & Wood, S. Agent-based model simulations of future changes in migration ows for Burina Faso. Global Environmental Change 21, 34–40 (2011). 20. Johnson, . T., Lampe, T. A. & Seichter, S. Calibration of an agent-based simulation model depicting a refugee camp scenario. In Proceedings of the 2009 Winter Simulation Conference (WSC), 1778–1786 (2009). 21. Soolowsi, J. A., Bans, C. M. & Hayes, . L. Modeling population displacement in the Syrian city of Aleppo. In Proceedings of the 2014 Winter Simulation Conference, 252–263 (2014). 22. labunde, A. & Willeens, F. Decision-maing in agent-based models of migration: State of the art and challenges. European Journal of Population 32, 73–97 (2016). 23. Castle, C. J. & Croos, A. T. Principles and concepts of agent-based modelling for developing geospatial simulations. Centre for Advanced Spatial Analysis, University College London (2006). 24. Croos, A., Castle, C. & Batty, M. ey challenges in agent-based modelling for geo-spatial simulation. Computers, Environment and Urban Systems 32, 417–430 (2008). 25. Borshchev, A. & Filippov, A. From system dynamics and discrete event to practical agent based modeling: easons, techniques, tools. In Proceedings of the 22nd International Conference of the System Dynamics Society, vol. 22 (2004). 26. Macal, C. M. & North, M. J. Tutorial on agent-based modelling and simulation. Journal of Simulation 4, 151–162 (2010). 27. Macal, C. M. Everything you need to now about agent-based modelling and simulation. Journal of Simulation 10, 144–156 (2016). 28. Elsenbroich, C. Explanation in agent-based modelling: Functions, causality or mechanisms? Journal of Articial Societies and Social Simulation 15, 1 (2012). 29. Epstein, J. M. Why model? Journal of Articial Societies and Social Simulation 11 (2008). 30. Entwisle, B. et al. Climate shocs and migration: An agent-based modeling approach. Population and Environment 1–25 (2016). 31. Hassani-Mahmooei, B. & Parris, B. W. Climate change and internal migration patterns in Bangladesh: An agent-based model. Environment and Development Economics 17, 763–780 (2012). 32. niveton, D. ., Smith, C. D. & Blac, . Emerging migration ows in a changing climate in dryland africa. Nature Climate Change 2, 444–447 (2012). 33. Anderson, J., Chaturvedi, A., Lengacher, D. & Cibulsis, M. Modeling the health of refugee camps: An agent-based computational approach. In 19th IEEE Symposium on Computer-Based Medical Systems (CBMS’06), 641–645 (2006). 34. Anderson, J., Chaturvedi, A. & Cibulsis, M. Simulation tools for developing policies for complex systems: Modeling the health and safety of refugee communities. Health care management science 10, 331–339 (2007). 35. Soolowsi, J. A. & Bans, C. M. A methodology for environment and agent development to model population displacement. In Proceedings of the 2014 Symposium on Agent Directed Simulation (2014). 36. Łate, M. M., izi, S. M. M. & Geller, A. Verication through calibration: An approach and a case study of a model of conict in syria. In Proceedings of the 2013 Winter Simulation Conference: Simulation: Making Decisions in a Complex World, 1649–1660 (2013). 37. Hattle, A., Yang, . S. & Zeng, S. Modeling the Syrian refugee crisis with agents and systems. e UMAP Journal 37, 195–213 (2016). 38. Collins, A. J. & Frydenlund, E. Agent-based modeling and strategic group formation: A refugee case study. Proceedings of the 2016 Winter Simulation Conference 1289–1300 (2016). 39. Lin, L., Carley, . M. & Cheng, S.-F. An agent-based approach to human migration movement. In Proceedings of the 2016 Winter Simulation Conference, 3510–3520 (2016). 40. Heath, B., Hill, . & Ciarallo, F. A survey of agent-based modelling practices (January 1998 to July 2008). Journal of Articial Societies and Social Simulation 12 (2009). 41. aleigh, C., Line, A., Hegre, H. & arlsen, J. Introducing ACLED-Armed Conict Location and Event Data. Journal of Peace Research 47, 651–660 (2010). 42. Lacher, W. Organized crime and conict in the Sahel-Sahara region. Carnegie Endowment for International Peace (2012). 43. Lecocq, B. et al. One hippopotamus and eight blind analysts: A multivocal analysis of the 2012 political crisis in the divided epublic of Mali. Review of African Political Economy 40, 343–357 (2013). 44. Turchin, P. Modeling social pressures toward political instability. Cliodynamics 4 (2013). 45. Turchin, P. Building nations aer conict. Nature 986–987 (2008). 46. Turchin, P. Long-term population cycles in human societies. Annals of the New York Academy of Sciences 1162, 1–17 (2009). 47. Turchin, P. D ynamics of political instability in the United States, 1780–2010. Journal of Peace Research 49, 577–591 (2012). 48. Boyce, M. & Vigaud-Walsh, F. ‘Youare either with us or against us’: Persecution and displacement in Bur undi. Refugees International (2015). 49. II. Burundi: A country on the edge. International efugee ights Initiative. Available at: http://www.refworld.org/ docid/57b6f9364.html (2016). 50. Yarnell, M. Central African epublic: e spotlight is gone, the crisis continues. Field report. efugees International (2015). 51. USCIF. Central African epublic. United States Commission on International eligious Freedom. Available at: http://www.uscirf. gov/sites/default/les/USCIF_A_2016_Tier1_2_CA.pdf (2016). 52. Hyndman, . J. & oehler, A. B. Another loo at measures of forecast accuracy. International Journal of Forecasting 22, 679–688 (2006). 53. Moore, W. H. & Shellman, S. M. efugee or internally displaced person? To where should one ee? Comparative Political Studies 39, 599–622 (2006). 54. UNHC. Global report. United Nations High Commissioner for efugees. Available at: http://www.unhcr.org/gr15/index.xml (2015). Acknowledgements We are grateful to Prof.Bastien Chopard, Prof.Peter Coveney, Dr.Moqi Groen-Xu and Prof.Simon Portegies Zwart for insightful discussions regarding this paper. Author Contributions D.S. and D.G. conceived the prediction approach and the experiments, D.B. advised on the design approach, D.S and D.G analysed the results, D.G. performed the sensitivity tests and comparison against naive models. All authors reviewed the manuscript. Content courtesy of Springer Nature, terms of use apply. Rights reserved www.nature.com/scientificreports/ 13 Scientific RepORtS | 7: 13377 | DOI:10.1038/s41598-017-13828-9 Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-017-13828-9. Competing Interests: e authors declare that they have no competing interests. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional aliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. e images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © e Author(s) 2017 Content courtesy of Springer Nature, terms of use apply. Rights reserved 1. 2. 3. 4. 5. 6. Terms and Conditions Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”). Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small- scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial. These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription (to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will apply. We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as detailed in the Privacy Policy. While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may not: use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access control; use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is otherwise unlawful; falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in writing; use bots or other automated methods to access the content or redirect messages override any security feature or exclusionary protocol; or share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal content. In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue, royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any other, institutional repository. These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved. To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law, including merchantability or fitness for any particular purpose. Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed from third parties. If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not expressly permitted by these Terms, please contact Springer Nature at onlineservice@springernature.com Available via license: CC BY 4.0 Content may be subject to copyright. SUPPLEMENTARY RESOURCE (1) Supplementary Material Data October 2017 Diana Suleimenova · David Bell · Derek Groen CITATIONS (90) REFERENCES (53) ... Despite several works in computational modeling that have tried to understand these dynamics from the perspective of destination countries (Suleimenova et al. 2017;Davis, Bhattachan et al. 2018;Asgary, Solis et al. 2016), we observe few works attempting to study these questions from the perspective of the origin country. And, if we take a step back, we realize that prior to understanding the choice of destination, we need to understand which individuals among the population in the affected region undergo forced migration. ... ... Searle and van Vuuren (2021) proposed a framework for modeling forced migration as a result of conflict events and used push and pull theory to determine the destination of migrants from Syria under that framework. Suleimenova et al. (2017) developed FLEE, a generalized simulation model that can be used to estimate the destination of migrants from a conflict-induced region. However, their work assumes that the people who want to migrate are already known and their model has to use this information as input. ... ... The result is shown in Figure 1d. The observed data is obtained from a previous study (Suleimenova et al. 2017). However, the data is associated with idiosyncrasies. ... A Generalizable Theory-Driven Agent-Based Framework to Study Conflict-Induced Forced Migration Article * Mar 2024 * Zakaria Mehrab * Logan Stundal * Srinivasan Venkatramanan * Madhav Marathe Large-scale population displacements arising from conflict-induced forced migration generate uncertainty and introduce several policy challenges. Addressing these concerns requires an interdisciplinary approach that integrates knowledge from both computational modeling and social sciences. We propose a generalized computational agent-based modeling framework grounded by Theory of Planned Behavior to model conflict-induced migration outflows within Ukraine during the start of that conflict in 2022. Existing migration modeling frameworks that attempt to address policy implications primarily focus on destination while leaving absent a generalized computational framework grounded by social theory focused on the conflict-induced region. We propose an agent-based framework utilizing a spatiotemporal gravity model and a Bi-threshold model over a Graph Dynamical System to update migration status of agents in conflict-induced regions at fine temporal and spatial granularity. This approach significantly outperforms previous work when examining the case of Russian invasion in Ukraine. Policy implications of the proposed framework are demonstrated by modeling the migration behavior of Ukrainian civilians attempting to flee from regions encircled by Russian forces. We also showcase the generalizability of the model by simulating a past conflict in Burundi, an alternative conflict setting. Results demonstrate the utility of the framework for assessing conflict-induced migration in varied settings as well as identifying vulnerable civilian populations. View Show abstract ... The proposed approach takes advantage of the searching abilities of populationbased meta-heuristic optimization approaches and the forecasting ability of agent-based simulation for the movements of asylum-seekers/unrecognized refugees. In contrast to single-objective optimization approaches converting the many-objective problem into a single-objective one, the employed multiobjective EAs (MOEAs), which search for the whole Pareto front of an MaOP, have been demonstrated to be more promising as they can help the search jump out of the local optima [18]. To show its performance, we conduct a case study of a recent South Sudan conflict and the experiments are based on popular MOEAs and a baseline algorithm. ... ... For example, the decision-maker is often not confident enough to specify the weights (i.e., to quantify relative importance) between the objectives. Moreover, a very recent study suggests that even if the weights can be accurately specified, the optimization approach searching for the whole Pareto front is more promising as it can help the search jump out of the local optima [18]. ... Many-Objective Simulation Optimization for Camp Location Problems in Humanitarian Logistics Article Full-text available * Sep 2024 * Yani Xue * Miqing Li * Hamid Arabnejad * Derek Groen Article Many-Objective Simulation Optimization for Camp Location Problems in Humanitarian Logistics Yani Xue 1,*, Miqing Li 2, Hamid Arabnejad 1, Diana Suleimenova 1, Alireza Jahani 1, Bernhard C. Geiger 3, Freek Boesjes 4, Anastasia Anagnostou 1, Simon J.E. Taylor 1, Xiaohui Liu 1, and Derek Groen 1,* 1 Department of Computer Science, Brunel University London, Uxbridge, United Kingdom 2 School of Computer Science, University of Birmingham, Birmingham, United Kingdom 3 Know-Center GmbH, Graz, Austria 4 Faculty of Geosciences, Utrecht University, Utrecht, The Netherlands * Correspondence: Yani Xue (Yani.Xue3@brunel.ac.uk); Derek Groen (Derek.Groen@brunel.ac.uk) Received: 10 March 2024 Accepted: 19 August 2024 Published: 26 September 2024 Abstract: Humanitarian organizations face a rising number of people fleeing violence or persecution, people who need their protection and support. When this support is given in the right locations, it can be timely, effective and cost-efficient. Successful refugee settlement planning not only considers the support needs of displaced people, but also local environmental conditions and available resources for ensuring survival and health. It is indeed very challenging to find optimal locations for establishing a new refugee camp that satisfy all these objectives. In this paper, we present a novel formulation of the facility location problem with a simulation-based evolutionary many-objective optimization approach to address this problem. We show how this approach, applied to migration simulations, can inform camp selection decisions by demonstrating it for a recent conflict in South Sudan. Our approach may be applicable to diverse humanitarian contexts, and the experimental results have shown it is capable of providing a set of solutions that effectively balance up to five objectives. View Show abstract ... To illustrate the toolchain and the main ideas behind model input verification, we use as a running example an agent-based simulation, called Flee [SBG17], designed for modeling displacement and migration patterns. Flee enables researchers to create simulations based on conflict and disaster scenarios, helping to predict how populations move in response to various crises. ... ... As already mentioned in § 3, Flee [SBG17] is a simulation tool designed for modeling displacement and migration patterns. It enables researchers to create simulations based on conflict and disaster scenarios, helping to predict how populations move in response to various crises. ... Model Input Verification of Large Scale Simulations Preprint Full-text available * Sep 2024 * Rumyana Neykova * Derek Groen Reliable simulations are critical for analyzing and understanding complex systems, but their accuracy depends on correct input data. Incorrect inputs such as invalid or out-of-range values, missing data, and format inconsistencies can cause simulation crashes or unnoticed result distortions, ultimately undermining the validity of the conclusions. This paper presents a methodology for verifying the validity of input data in simulations, a process we term model input verification (MIV). We implement this approach in FabGuard, a toolset that uses established data schema and validation tools for the specific needs of simulation modeling. We introduce a formalism for categorizing MIV patterns and offer a streamlined verification pipeline that integrates into existing simulation workflows. FabGuard's applicability is demonstrated across three diverse domains: conflict-driven migration, disaster evacuation, and disease spread models. We also explore the use of Large Language Models (LLMs) for automating constraint generation and inference. In a case study with a migration simulation, LLMs not only correctly inferred 22 out of 23 developer-defined constraints, but also identified errors in existing constraints and proposed new, valid constraints. Our evaluation demonstrates that MIV is feasible on large datasets, with FabGuard efficiently processing 12,000 input files in 140 seconds and maintaining consistent performance across varying file sizes. View Show abstract ... Within the same area of application, another type of migration is gaining attention. That is, forced migration as a result of an armed conflict(Hébert, Perez, and Harati 2018;Perez Estrada, Groen, and Ramirez-Marquez 2017;Suleimenova, Bell, and Groen 2017), where modelling through agents have allowed researchers to simulate individual's decision reasoning process in order to decide where to relocate after being forced to move. ... Geography Compass Big Data (R)evolution in Geography: Complexity Modelling in the Last Two Decades Article Full-text available * Nov 2024 * Liliana Pérez * Raja Sengupta The use of data and statistics along with computational systems heralded the beginning of a quantitative revolution in Geography. Use of simulation models (Cellular Automata and Agent-Based Models) followed in the late 1990s, with ontology and epistemology of complexity theory and modelling being defined a little less than two decades ago. We are, however, entering a new era where sensors regularly collect and update large amounts of spatio-temporal data. We define this 'Big Data' as geo-located data collected in sufficiently high volume (exceeding storage capacities of the largest personal hard drives currently available), that is updated at least daily, from a variety of sources in different formats, often without recourse to verification of its accuracy. We then identify the exponential growth in the use of complexity simulation models in the past two decades via an extensive literature review (broken down by application area), but also notice a recent slowdown. Further, a gap in the utilisation of Big Data by modellers to calibrate and validate their models is noted, which we attribute to data availability issues. We contend that Big Data can significantly boost simulation modelling, if certain constraints and issues are managed properly. View Show abstract ... 40 Certainly, a more agnostic data-driven approach can enhance modeling and forecasting, being free from assumptions about variable relations and interaction terms. Bayesian networks, 41 generalized models and simulation frameworks, 42 time series forecasting 43 and hypothesis testing through regression 27,44 are some of the methods used. However, these models only capture association patterns and are thus not guaranteed to capture true causal relations between the involved drivers, impeding an accurate attribution of cause and effect relationships. ... Causal discovery reveals complex patterns of drought-induced displacement Article Full-text available * Aug 2024 * José María Tárraga * Eva Sevillano-Marco * Jordi Muñoz * Gustau Camps-Valls The increasing frequency and severity of droughts present a significant risk to vulnerable regions of the globe, potentially leading to substantial human displacement in extreme situations. Drought-induced displacement is a complex and multifaceted issue that can perpetuate cycles of poverty, exacerbate food and water scarcity, and reinforce socio-economic inequalities. However, our understanding of human mobility in drought scenarios is currently limited, inhibiting accurate predictions and effective policy responses. Drought-induced displacement is driven by numerous factors and identifying its key drivers, causal-effect lags, and consequential effects is often challenging, typically relying on mechanistic models and qualitative assumptions. This paper presents a novel, data-driven methodology, grounded in causal discovery, to retrieve the drivers of drought-induced displacement within Somalia from 2016 to 2023. Our model exposes the intertwined vulnerabilities and the leading times that connect drought impacts, water and food security systems along with episodes of violent conflict, emphasizing that causal mechanisms change across districts. These findings pave the way for the development of algorithms with the ability to learn from human mobility data, enhancing anticipatory action, policy formulation, and humanitarian aid. View Show abstract ... Third, there is potential for the estimation of socioeconomic indicators using alternative data sources and advanced analytics, and this potential is worth investigating and testing. Fourth, given the complexities of displacement prediction in general (Suleimenova et al., 2017;Huynh and Basu, 2020;Leasure et al., 2022;Pham and Luengo-Oroz, 2023), 15 including its intersections with environmental modeling in humanitarian contexts 16 and conflict prediction, 17 any new drive for the integration of socioeconomic estimates into predictive models of displacement could prove premature, as well as giving rise to new data responsibility concerns. 18 Given these main outcomes from discussions, as well as the themes that emerged across topic areas, particularly the emphasis on complementarity to traditional approaches, we have re-organized the potential contribution of innovation and data science to socioeconomic data on forced displacement identified by the workshop, emphasizing operational relevance and feasibility, by stage of the datacollection-to-use pipeline. ... Accelerating and enhancing the generation of socioeconomic data to inform forced displacement policy and response Article Full-text available * Dec 2023 * Paddy Brock * Harriet Kasidi Mugera There are now an estimated 114 million forcibly displaced people worldwide, some 88% of whom are in low- and middle-income countries. For governments and international organizations to design effective policies and responses, they require comparable and accessible socioeconomic data on those affected by forced displacement, including host communities. Such data is required to understand needs, as well as interactions between complex drivers of displacement and barriers to durable solutions. However, high-quality data of this kind takes time to collect and is costly. Can the ever-increasing volume of open data and evolving innovative techniques accelerate and enhance its generation? Are there applications of alternative data sources, advanced statistics, and machine-learning that could be adapted for forced displacement settings, considering their specific legal and ethical dimensions? As a catalytic bridge between the World Bank and UNHCR, the Joint Data Center on Forced Displacement convened a workshop to answer these questions. This paper summarizes the emergent messages from the workshop and recommendations for future areas of focus and ways forward for the community of practice on socioeconomic data on forced displacement. Three recommended areas of future focus are: enhancing and optimizing household survey sampling approaches; estimating forced displacement socioeconomic indicators from alternative data sources; and amplifying data accessibility and discoverability. Three key features of the recommended approach are: strong complementarity with the existing data-collection-to-use-pipeline; data responsibility built-in and tailored to forced displacement contexts; and iterative assessment of operational relevance to ensure continuous focus on improving outcomes for those affected by forced displacement. View Show abstract Capacitated Mobile Facility Location Problem with Mobile Demand: Efficient Relief Aid Provision to En Route Refugees Article * Jul 2024 * OMEGA-INT J MANAGE S * Amirreza Pashapour * Dilek Günneç * F. Sibel Salman * Eda Yücel As a humanity crisis, the tragedy of forced displacement entails relief aid distribution efforts among en route refugee to alleviate their migration hardships. This study aims to assist humanitarian organizations in cost-efficiently optimizing the logistics of capacitated mobile facilities utilized to deliver relief aid to transiting refugees in a multi-period setting. The problem is referred to as the Capacitated Mobile Facility Location Problem with Mobile Demands (CMFLP-MD). In CMFLP-MD, refugee groups follow specific paths, and meanwhile, they receive relief aid at least once every fixed number of consecutive periods, maintaining continuity of service. To this end, the overall costs associated with capacitated mobile facilities, including fixed, service provision, and relocation costs, are minimized. We formulate a mixed integer linear programming (MILP) model and propose two solution methods to solve this complex problem: an accelerated Benders decomposition approach as an exact solution method and a matheuristic algorithm that relies on an enhanced fix-and-optimize agenda. We evaluate our methodologies by designing realistic instances based on the Honduras migration crisis that commenced in 2018. Our numerical results reveal that the accelerated Benders decomposition excels MILP with a 46% run time improvement on average while acquiring solutions at least as good as the MILP across all instances. Moreover, our matheuristic acquires high-quality solutions with a 2.4% average gap compared to best-incumbents rapidly. An in-depth exploration of the solution properties underscores the robustness of our relief distribution plans under varying migration circumstances. Across several metrics, our sensitivity analyses also highlight the managerial advantages of implementing CMFLP MD solutions. View Show abstract Flee 3: Flexible agent-based simulation for forced migration Article * Jun 2024 * Maziar Ghorbani * Diana Suleimenova * Alireza Jahani * Derek Groen View The eco-evolutionary dynamics of strategic species Article Full-text available * Apr 2024 * Sourav Roy * Subrata Ghosh * Arindam Saha * Dibakar Ghosh Much research has in recent years been devoted to better our understanding of the intricate relationships between ecology and the evolutionary success of species. These explorations have often focused on understanding the complex interplay among ecological factors and evolutionary rhythms of the species in various environments. Central to these studies is the concept of the survival of the fittest, proposed by Charles Darwin, where evolutionary circumstances, often portrayed as social dilemmas, favour the welfare of self-interested over others. To further advance this line of research, we here develop a theoretical framework that features three interconnected traits in an evolutionary setting, namely: prey, predator and parasite, each adopting distinct strategies akin to a social dilemma and resembling a Rock-Paper-Scissors scenario. These traits, which we term strategic species, adhere to the eco-evolutionary game dynamics. We further extend our analysis by conducting a sensitivity assessment of the system’s payoff parameters using the Sobol indices. View Show abstract An agent-based framework to study forced migration: A case study of Ukraine Article * Mar 2024 * Zakaria Mehrab * Logan Stundal * Srinivasan Venkatramanan * Madhav Marathe The ongoing Russian aggression against Ukraine has forced over eight million people to migrate out of Ukraine. Understanding the dynamics of forced migration is essential for policy-making and for delivering humanitarian assistance. Existing work is hindered by a reliance on observational data which is only available well after the fact. In this work, we study the efficacy of a data-driven agent-based framework motivated by social and behavioral theory in predicting outflow of migrants as a result of conflict events during the initial phase of the Ukraine war. We discuss policy use cases for the proposed framework by demonstrating how it can leverage refugee demographic details to answer pressing policy questions. We also show how to incorporate conflict forecast scenarios to predict future conflict-induced migration flows. Detailed future migration estimates across various conflict scenarios can both help to reduce policymaker uncertainty and improve allocation and staging of limited humanitarian resources in crisis settings. View Show abstract Show more Evaluation of existing migration forecasting methods and models. Report for the Migration Advisory Committee Technical Report Full-text available * Oct 2015 * George Disney * Arkadiusz Wiśniowski * Jonathan J. Forster * Jakub Bijak This report includes a detailed review of existing migration forecasting methods and models and an exercise which compares the forecasts from the models identified with actual historic migration flows. View Show abstract A Serious Video Game To Support Decision Making On Refugee Aid Deployment Policy Article Full-text available * Dec 2017 * Luis Eduardo Estrada * Derek Groen * Jose Emmanuel Ramirez-Marquez The success of refugee support operations depends on the ability of humanitarian organizations and governments to deploy aid effectively. These operations require that decisions on resource allocation are made as quickly as possible in order to respond to urgent crises and, by anticipating future developments, remain adequate as the situation evolves. Agent-based modeling and simulation has been used to understand the progression of past refugee crises, as well as a way to predict how new ones will unfold. In this work, we tackle the problem of refugee aid deployment as a variant of the Robust Facility Location Problem (RFLP). We present a serious video game that functions as an interface for an agent-based simulation run with data from past refugee crises. Having obtained good approximate solutions to the RFLP by implementing a game that frames the problem as a puzzle, we adapted its mechanics and interface to correspond to refugee situations. The game is intended to be played by both subject matter experts and the general public, as a way to crowd-source effective courses of action in these situations. View Show abstract Everything you need to know about agent-based modelling and simulation Article Full-text available * May 2016 * J Simulat * C. M. Macal This paper addresses the background and current state of the field of agent-based modelling and simulation (ABMS). It revisits the issue of ABMS represents as a new development, considering the extremes of being an overhyped fad, doomed to disappear, or a revolutionary development, shifting fundamental paradigms of how research is conducted. This paper identifies key ABMS resources, publications, and communities. It also proposes several complementary definitions for ABMS, based on practice, intended to establish a common vocabulary for understanding ABMS, which seems to be lacking. It concludes by suggesting research challenges for ABMS to advance and realize its potential in the coming years. View Show abstract Decision-Making in Agent-Based Models of Migration: State of the Art and Challenges Article Full-text available * Feb 2016 * EUR J POPUL * Anna Klabunde * Frans Willekens We review agent-based models (ABM) of human migration with respect to their decision-making rules. The most prominent behavioural theories used as decision rules are the random utility theory, as implemented in the discrete choice model, and the theory of planned behaviour. We identify the critical choices that must be made in developing an ABM, namely the modelling of decision processes and social networks. We also discuss two challenges that hamper the widespread use of ABM in the study of migration and, more broadly, demography and the social sciences: (a) the choice and the operationalisation of a behavioural theory (decision-making and social interaction) and (b) the selection of empirical evidence to validate the model. We offer advice on how these challenges might be overcome. View Show abstract Climate Shocks and Migration: An Agent-Based Modeling Approach Article Full-text available * Sep 2016 * POPUL ENVIRON * Barbara Entwisle * Nathalie E. Williams * Ashton M. Verdery * Aree Jampaklay This is a study of migration responses to climate shocks. We construct an agent-based model that incorporates dynamic linkages between demographic behaviors, such as migration, marriage, and births, and agriculture and land use, which depend on rainfall patterns. The rules and parameterization of our model are empirically derived from qualitative and quantitative analyses of a well-studied demographic field site, Nang Rong district, northeast Thailand. With this model, we simulate patterns of migration under four weather regimes in a rice economy: (1) a reference, “normal” scenario; (2) 7 years of unusually wet weather; (3) 7 years of unusually dry weather; and (4) 7 years of extremely variable weather. Results show relatively small impacts on migration. Experiments with the model show that existing high migration rates and strong selection factors, which are unaffected by climate change, are likely responsible for the weak migration response. View Show abstract Organized Crime and Conflict in the Sahel-Sahara Region Chapter * Apr 2013 * Wolfram Lacher View Modeling Social Pressures Toward Political Instability Article * Dec 2013 * Peter Turchin View An agent-based approach to human migration movement Conference Paper * Dec 2016 * Larry Lin * Kathleen M Carley * Shih-Fen Cheng View Agent-based modeling and strategic group formation: A refugee case study Conference Paper * Dec 2016 * Andrew Collins * Erika Frydenlund View Simulating Refugee Movements: Where Would You Go? Article * Dec 2016 * Derek Groen The challenge of understanding refugee movements is huge and affects countries worldwide on a daily basis. Yet, in terms of simulation, the challenge appears to have been largely ignored. I argue that we as researchers can, and should, harness our computational skills to better understand and predict refugee movements. I reflect on the computational challenges of modelling refugees, and present a simulation case study example focused on the Northern Mali Conflict in 2012. Compared to UNHCR data, the simulation predicts fewer refugees moving towards Mauritania, and more refugees moving towards Niger. This outcome aligns with UNHCR reports, which mention that unregistered refugees were known to reside outside of the official camps, though further investigations are required to rule out competing theories. View Show abstract Show more RECOMMENDED PUBLICATIONS Discover more Thesis Full-text available HUMANITARIAN ACTION & LEVELS OF VIOLENCE IN NORTH KIVU December 2016 * Paul Daragh Muldoon This paper tests theories which argue that humanitarian aid often negatively impacts the violent conflicts it responds to. Using the territories of North Kivu in the Democratic Republic of Congo as a series of case studies combined with historical statistical datasets for that region, I show that the available evidence fails to support such a position. Instead, I argue that, not only do aid sites ... [Show full abstract] mitigate many of the negative effects of violence and related displacement, they also offer a place of refuge for civilians during periods of escalating violence. View full-text Article Full-text available HUMANITARIAN ORGANIZATIONS IN TAJIKISTAN AND THE COORDINATION OF AID TO DISPLACED AFGHANS IN NO MAN'... June 2005 · Journal of Refugee Studies * Indra Overland This article examines the coordination of humanitarian aid to displaced Afghans on the border between Afghanistan and Tajikistan from 2000 to 2002, in order to draw lessons that may be useful in connection with the return of the remaining displaced Afghans and other groups of refugees and IDPs. It examines the roles of the various organizations involved in the humanitarian operation, focusing on ... [Show full abstract] UNHCR and its position as lead agency. The case of the displaced Afghans on the border between Afghanistan and Tajikistan highlights some problematic consequences of locating coordinating functions in sectoral lead agencies, and some advantages of neutral coordination bodies such as OCHA. It also demonstrates the importance of a balance between information gathering, processing and sharing, as well as the interconnections between coordination by command and coordination by consensus. View full-text Article Full-text available LOCATING YOUNG REFUGEES HISTORICALLY: ATTENDING TO AGE POSITION IN HUMANITARIANISM April 2014 · European Journal of Development Research * Jason Hart Nowadays humanitarian organisations are often keen to engage young displaced people in programmatic efforts. In such efforts attention is commonly paid to the impact of the social dynamics of gender. However, similar consideration of processes associated with age has been less apparent. This article explains the importance of attending to the ‘age position’ of young refugees from two interrelated ... [Show full abstract] perspectives: first, as a means to comprehend the forces that inform the expression of particular needs and aspirations by young people, and second, in order to grasp the historicity of their lives and of the larger displaced population. The article then moves on to offer a conceptual framework for investigating age position. The notion of ‘generation’ is central to this framework. Four distinct meanings of generation are identified and their application explored through reference to findings from research conducted in a Palestinian refugee camp in Jordan.Les organisations humanitaires sont aujourd’hui souvent désireuses de faire participer les jeunes déplacés à des initiatives de programmes. Dans ces programmes, l’attention se porte généralement sur l'impact de la dynamique sociale entre les sexes. On semble par contre accorder moins d’attention aux dynamiques liées à l'âge. Cet article explique, à partir de deux perspectives connexes, l'importance d’aborder la question du «rang d’âge» des jeunes réfugiés. Il s’agit, tout d'abord, de comprendre les forces qui conditionnent l'expression des besoins et aspirations particuliers des jeunes et, puis de mieux appréhender l'historicité de leur vie et de celle de l'ensemble des réfugiés. L’article poursuit en proposant un cadre conceptuel pour examiner l’importance du rang d’âge. La notion de «génération» occupe une place centrale dans ce cadre. Nous identifions quatre sens distincts de «génération» et examinons leur application en mobilisant des résultats d’études menées dans un camp de réfugiés palestiniens en Jordanie. View full-text Article WHO ARE NEEDED, THIEVES OR DOCTORS: ARMED CIVIL CONFLICTS IN DARFUR AND IMPACTS ON EDUCATION IN REFU... July 2011 * Issam A.W. Mohamed The study presented here depends on a field survey of refugees' camps in war strifed Darfur region. The data are genuine from people and children. The expelling of NGOs from the region was a moral shock to the whole world, but in Darfur, it was a humanitarian catastrophe affecting the population. In this paper, I surveyed and analyzed data on what I consider the most important factor which are ... [Show full abstract] children and establish education as the base of the arguments introduced here. A case study is introduced which is Attash refugees or displaced camp were produced. Data were collected by field questionnaire of the inhabitants of the camp that concentrated on the education backgrounds and the available facilities to educate children. That is besides collected data on socioeconomic and demographic conditions. The results and conclusions emphasize that such facilities are inadequate or non-existent. Malnutrition, inadequate homes and sanitation facilities are common. Very few job opportunities are available and there is no land for the displaced people to practice their traditional profession. Governmental supporting systems are not available or do not exist. Drinking water facilities are insufficient. All indices of human development are very low. Thus, no sane person can reject foreign humanitarian aid. The title for the paper introduced here is that Doctors or Thieves. For the first, they are not sufficiently available and if there is any, they don’t have medicines. Of the second, the current conditions only generate thieves in all possible categories who will infest the whole region and country; generations of young people who socially and morally are failures not provided or educated to participate in building the society. Besides my conclusions on impacts on education, I hereby profess that what happens on the camps generates only further economic destitute and entrenching social hatred in the Sudanese nation. Read more Last Updated: 22 Oct 2024 Discover the world's research Join ResearchGate to find the people and research you need to help your work. Join for free ResearchGate iOS App Get it from the App Store now. Install Keep up with your stats and more Access scientific knowledge from anywhere or Discover by subject area * Recruit researchers * Join for free * Login Email Tip: Most researchers use their institutional email address as their ResearchGate login PasswordForgot password? Keep me logged in Log in or Continue with Google Welcome back! Please log in. Email · Hint Tip: Most researchers use their institutional email address as their ResearchGate login PasswordForgot password? Keep me logged in Log in or Continue with Google No account? Sign up Company About us News Careers Support Help Center Business solutions Advertising Recruiting © 2008-2024 ResearchGate GmbH. All rights reserved. * Terms * Privacy * Copyright * Imprint * Consent preferences