Enhanced rainfall
prediction in Gujarat, India using advanced machine learning models
Shaikh A.F., Kharat P.V., Pujari A.B., Gunaware P.D. and Darade M.M.
Disaster Advances; Vol. 18(12); 51-58;
doi: https://doi.org/10.25303/1812da051058; (2025)
Abstract
Precise prediction of precipitation is essential for efficient management of water
resources, planning of agriculture and readiness for disasters, particularly in
areas like Gujarat, India, where climate fluctuations are common. This study uses
cutting-edge machine learning methods such as XGBoost and CatBoost, to improve rainfall
forecasts made from historical rainfall data. Important metrics including R-squared
(R²), Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) are used to compare
and to assess the performance of these models.
In training, testing and validation datasets, CatBoost consistently outperforms
XGBoost in terms of prediction accuracy, as evidenced by greater R2 values and lower
RMSE and MAE values. These results imply that CatBoost is a better option for rainfall
prediction jobs as it is more adept at identifying patterns and trends in the rainfall
data. The study's outcomes have significant implications for Gujarat's ability to
predict rainfall accurately. Improved predictions can aid in better planning for
water storage and distribution, optimize agricultural schedules and enhance flood
management strategies.
