Department of Conflict Resolution Studies Theses and Dissertations

Copyright Statement

All rights reserved. This publication is intended for use solely by faculty, students, and staff of Nova Southeastern University. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, now known or later developed, including but not limited to photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author or the publisher.

Date of Award

1-1-2026

Document Type

Dissertation - NSU Access Only

Degree Name

Doctor of Philosophy in Conflict Analysis and Resolution

First Advisor

Elena Bastidas

Keywords

Artificial Intelligence; Humanitarian-Aid; Internally Displaced Persons (IDP); Predictive-Analytics; Syria; Systems-Modeling

Abstract

This dissertation presents an embedded quantitative case study of internal displacement in Syria (2011-2019) and an applied framework for anticipatory humanitarian aid. The study asks whether routinely collected secondary data can explain internally displaced person (IDP) movements and support predictive, auditable decisions about where and when to pre-position assistance. A multidimensional (secondary) dataset is assembled from established humanitarian and conflict information streams, linking displacement observations to conflict intensity and event dynamics, access constraints, and selected socioeconomic pressures captured in operational reporting. Chapter 4 constructs an analytics-to-operations pipeline that distinguishes displacement flows from stock, prioritizes calibration, and evaluates forecasts with rolling-origin validation to reduce temporal leakage. Five model families are developed for complementary decision support: time-series forecasting (ETS/ARIMA) for near-term magnitude envelopes, supervised learning for district-level surge risk and ranking, a transparent gravity baseline for conservative origin-destination plausibility, spatial-temporal Bayesian networks for probabilistic dependency structure, and Agent-based modeling for scenario testing under corridor and capacity constraints. Predictions are operationalized through recall-favoring triggers, a Priority Index that generates a ranked decision list, and sector calculators that translate forecasted arrivals into feasible delivery plans. Chapter 5 interprets implications for coordination, governance, and do-no-harm safeguards and demonstrates how an AI assistant can enable non-specialists to submit structured field inputs and receive plain-language, traceable recommendations. Limitations include reporting bias, missingness, and the absence of prospective pilots. The dissertation concludes that transparent, calibrated predictive analytics can improve lead time and targeting compared with legacy, reactive intuition-driven allocation in complex emergencies.

Download
To access this thesis/dissertation you must have a valid nova.edu OR mynsu.nova.edu email address and create an account for NSUWorks.

Free My Thesis

If you are the author of this work and would like to grant permission to make it openly accessible to all, please click the Free My Thesis button.

  Contact Author

Share

COinS