# Forecasting Thailand’s mobility trends using Feature Engineered XGBoost for pandemic crisis movement management

## TL;DR

This paper uses machine learning to forecast mobility trends in Thailand during the pandemic, showing how movement patterns can inform public health decisions.

## Contribution

The study introduces Feature Engineered XGBoost as a novel and effective model for forecasting mobility trends during public health crises.

## Key findings

- Feature Engineered XGBoost outperformed Prophet and ARIMA in forecasting mobility trends.
- Mobility patterns were found to be significantly related to changes in COVID-19 case numbers during lockdown phases.
- Machine learning models can effectively support data-driven public health interventions.

## Abstract

The COVID-19 pandemic significantly disrupted global mobility patterns, with widespread population movement playing a key role in the transmission of the virus. In such a situation, Google introduced the Community Mobility Reports, which use anonymized and aggregated location data to monitor changes in movement across various location categories. These mobility trends provide important insights that help inform timely public health interventions and support data-driven decisions during and after the pandemic. This study aims to forecast human mobility trends in Thailand during the COVID-19 pandemic using data from Google’s reports. Three forecasting models were applied: Facebook Prophet, ARIMA, and Feature Engineered XGBoost. The Granger Causality Test was used to examine the relationship between mobility patterns and COVID-19 case numbers across different phases of lockdown. The results indicated that Feature Engineered XGBoost demonstrated the highest overall accuracy in forecasting mobility trends across all six location categories. In conclusion, this study demonstrates the effectiveness of machine learning models in forecasting mobility movement across various location types while public health restrictions have been implemented. This underscores the importance of understanding mobility patterns as a key factor in disease transmission. The insights gained from this analysis can help formulate strategic and targeted mobility management policies and public health responses for future outbreaks, ultimately helping to contain the spread of disease more effectively.