Can LLM-assisted regularization increase forecast accuracy for migration flows in low data regimes?

Nathaniel T. Hindman; Fabricio Murai

cs.LGOct 5, 2026

Can LLM-assisted regularization increase forecast accuracy for migration flows in low data regimes?

Authors: Nathaniel T. Hindman, Fabricio Murai

Organizations: Artificial Intelligence Worcester Polytechnic Institute Worcester, Massachusetts · Data Science Worcester Polytechnic Institute Worcester, Massachusetts

Abstract

Predicting migration flows remains a significant challenge for traditional gravity-based forecasting models, which primarily rely on structured socio-economic indicators such as economic disparity, political stability, and geographic distance. This work investigates whether Large Language Models (LLMs) can improve migration forecasting by extracting contextual migration-related signals from news articles and incorporating them into a weighted Lasso forecasting framework through feature-specific regularization penalties. The proposed framework uses hierarchical LLM inference pipelines to classify migration-related push--pull signals from news data and evaluates the resulting forecasting performance across multiple migration corridors between November 2021 and November 2022, including Mexico--United States, Ukraine--Poland, and Syria--Turkey. Experimental results showed mixed performance across migration corridors and modeling strategies, and no single regularization approach consistently outperformed the others across all experiments. The best-performing Mexico configuration, which consisted of a gravity-based model augmented with the proposed push--pull ratios, achieved a Mean Absolute Percentage Error (MAPE) of 17.15%, while the strongest Syria configuration achieved a MAPE of 29.29% using Direct LLM-Lasso. For Ukraine, the best-performing configuration used LLM-Assisted Regularization (AR) and achieved a MAPE of 41.05%. Overall, the results suggest that contextual article-derived features and LLM-guided regularization can improve migration forecasting under certain conditions, although migration corridor characteristics, article volume, and hyperparameter configuration strongly influenced performance.

Figures & tables

Corridor	Min	Max	Mean	Std. Dev.
Mexico–US	8,718	102,939	46,194	23,840
Ukraine–Poland	998	163,937	18,628	30,839
Syria–Turkey	1,959	17,395	5,821	3,325

TABLE I: Summary statistics for migration-flow datasets across evaluated migration corridors.

Fig. 1: SuLuSa Classification across steps

Fig. 2: SuLu Classification across steps

Mode	Push	Pull	Mixed	Unclear
Qwen SuSa	162	24	0	195
Qwen SuLu	126	35	1	219
Qwen SuLuSa	157	60	1	163
Mistral SuSa	150	40	15	172

TABLE II: Inference Pipeline Performance at Temperature

\tau=0.8

	Gravity	Best	Direct	LLM-AR
	Gravity	Gravity-PP	LLM-Lasso	LLM-AR
Mexico	17.44	17.15	23.48	23.92
Ukraine	61.66	42.10	43.61	41.05
Syria	31.44	32.74	29.29	32.26

TABLE III: Direct LLM-Lasso performance compared to control and best gravity-based forecasting configurations (MAPE %)

Fig. 3: Box plot comparison of absolute percentage error distributions across migration forecasting models for Mexico, Syria, and Ukraine under varying article counts and regularization settings. Lower values indicate improved predictive accuracy.

Fig. 4: N-comparison across Models

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

Temp.	Mode	Push	Pull	Mixed	Unclear
0.3	Qwen SuSa	152	27	2	200
	Qwen SuLu	124	46	0	211
	Qwen SuLuSa	164	54	0	163
	Mistral SuSa	160	42	4	168
0.8	Qwen SuSa	162	24	0	195
	Qwen SuLu	126	35	1	219

Appendix

TABLE IV: Temperature sensitivity results.

Temperature	Push	Pull	Mixed	Unclear
0.1	165	55	1	160
0.3	164	54	0	163
0.5	173	46	4	158
0.8	157	60	1	163
0.9	159	48	2	170
1.5	175	41	6	159

Appendix

TABLE V: Inference Performance for mode SuLuSa across temperatures

Forecasting Results
Migration Corridor	Best Model	$\boldsymbol{\lambda}$	$N$ Articles
Mexico	No $w_{j}$	0.003	1,245
Ukraine	LLM-Assisted Reg. (AR)	0.003	100
Syria	Direct Lasso	0.005	882

Appendix

TABLE VI: Best-performing forecasting configurations across migration corridors.

Hyper-Parameter	Evaluated Values
Lasso Regularization Strength ( $\lambda$ )	$7\times 10^{-4}$ , $9\times 10^{-4}$ , $1\times 10^{-3}$ ,
	$3\times 10^{-3}$ , $5\times 10^{-3}$ , $6\times 10^{-3}$ ,
	$7\times 10^{-3}$ , $8\times 10^{-3}$
Primary Experimental Values	$\{1,3,5,8\}\times 10^{-3}$

Appendix

TABLE VII: Evaluated Lasso regularization strengths used throughout forecasting experiments.

Explore similar work

Do LLMs Know Your Neighborhood? Auditing LLM Priors for Neighborhood-Level Mobility Prediction and Structural Alignment
Aug 31, 2026Saad Mohammad Abrar, Eesha Kurella, Arnav Dadarya +3Mobility Computational Social Science
University of Maryland USA
ARMove: Learning to Predict Human Mobility through Agentic Reasoning
Apr 19, 2026Chuyue Wang, Jie Feng, Yuxi Wu +2Mobility Human Motion Prediction
Tencent · Zhongguancun Academy · Alibaba Group +1
Towards Efficient and Evidence-grounded Mobility Prediction with LLM-Driven Agent
Jun 3, 2026Linyao Chen, Qinlao Zhao, Zechen Li +7Mobility
The University of Tokyo · Huazhong University of Science and Technology · University of New South Wales, Sydney +3

Aug 31, 2026, cs.LGJ/K move · Enter open · S save

Do LLMs Know Your Neighborhood? Auditing LLM Priors for Neighborhood-Level Mobility Prediction and Structural Alignment

Saad Mohammad Abrar, Eesha Kurella, Arnav Dadarya, Naman Awasthi+2

University of Maryland USA

Mobility Computational Social Science

Open paper AskPDF arXiv

Human mobility is central to urban planning, transportation, public health, and emergency response, yet fine-grained trajectory data are often proprietary, restricted, and privacy-sensitive. Large language models (LLMs) offer a potential alternative by generating plausible mobility traces and predicting individual movement, but their ability to infer aggregate neighborhood-level mobility remains unclear. We evaluate zero-shot LLMs on Census Block Group-level mobility prediction across four U.S. metropolitan areas using anonymized Cuebiq data to construct point-level, trajectory-level, and temporal mobility outcomes, paired with sociodemographic and built-environment predictors. We compare LLM predictions with supervised baselines and introduce a directional alignment analysis to test whether LLM-implied predictor effects agree with empirical OLS and Jonckheere-Terpstra trends. Supervised models achieve 0.580 average accuracy, compared with 0.435 for the best LLM, with spatial extent outcomes showing the strongest predictability but also the largest LLM-baseline gaps. Directional analysis shows that LLMs often rely on coarse, stable predictor-level priors that remain similar across outcomes and cities, including asymmetric treatment of protected-group predictors. Overall, LLMs can partially recover aggregate mobility patterns from urban context, but their predictions should not be treated as structurally grounded without auditing empirical alignment and potential bias.

Temperature	Push	Pull	Mixed	Unclear
0.1	165	55	1	160
0.3	164	54	0	163
0.5	173	46	4	158
0.8	157	60	1	163
0.9	159	48	2	170
1.5	175	41	6	159

Temperature	Push	Pull	Mixed	Unclear
0.1	165	55	1	160
0.3	164	54	0	163
0.5	173	46	4	158
0.8	157	60	1	163
0.9	159	48	2	170
1.5	175	41	6	159

Abstract

Explore similar work

Do LLMs Know Your Neighborhood? Auditing LLM Priors for Neighborhood-Level Mobility Prediction and Structural Alignment

ARMove: Learning to Predict Human Mobility through Agentic Reasoning

Towards Efficient and Evidence-grounded Mobility Prediction with LLM-Driven Agent

Temperature	Push	Pull	Mixed	Unclear
0.1	165	55	1	160
0.3	164	54	0	163
0.5	173	46	4	158
0.8	157	60	1	163
0.9	159	48	2	170
1.5	175	41	6	159