Python ARIMA Code for Thailand COVID-19 Mobility Forecasting

A 6.5 KB text file containing Python code for an ARIMA model. The code was developed by Aritath Siraphatwongkorn to forecast human mobility trends in Thailand during the COVID-19 pandemic, using data from Google Community Mobility Reports. The study compared ARIMA against Facebook Prophet and Feature Engineered XGBoost models.

Use Cases

• Implement the provided ARIMA model code to forecast mobility trends for six location categories in Thailand.
• Replicate the study's methodology for examining relationships between mobility patterns and COVID-19 case numbers using the Granger Causality Test.
• Use the code as a benchmark to compare against other forecasting models like Facebook Prophet or Feature Engineered XGBoost for time series analysis.

Strengths

• Code is focused on a specific, well-defined application: forecasting mobility during Thailand's COVID-19 pandemic.
• The study methodology is explicitly described, comparing three distinct forecasting models (ARIMA, Prophet, XGBoost).
• Data source is clearly identified as Google's Community Mobility Reports for six location categories.

Limitations

• The dataset is extremely small (6.5 KB), containing only code and no actual mobility or case data.
• Users must source the underlying Google mobility data and COVID-19 case numbers separately to execute the code.
• The code's performance and applicability are tied to the specific temporal and geographic context (Thailand, pandemic period).

Provenance

Source

figshare, authored by Aritath Siraphatwongkorn.

Collection Method

Code developed for a research study comparing forecasting models.

Time Range

COVID-19 pandemic period in Thailand.

Freshness

Last updated in March 2026.

Geography

Thailand.