Paper Key : IRJ************770
Author: Devasheesh Krishan
Date Published: 02 Oct 2023
Abstract
Accurate precipitation forecasting is of paramount importance in various fields, including agriculture, hydrology, environmental engineering and disaster management. This study investigates the suitability and performance of different regression models in predicting precipitation based on meteorological variables. Seven meteorological factors, including mean PBLH (Planetary Boundary Layer Height), wind speed, humidity, evapo-transpiration,, NDVI (Normalized Difference Vegetation Index), surface radiation and temperature, are considered as potential predictors of rainfall in this study.The objective is to identify the most effective regression model for quantifying the dependency of precipitation on these meteorological parameters. To achieve this, a comprehensive comparative analysis of regression models available in the PyCaret framework is conducted. Models such as Linear Regression, Lasso Regression, Random Forest Regression, K-Neighbors Regression and XGradient Boosting Regression, among others, are evaluated and compared. The best model is chosen based on certain parameters. The assessment includes metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), R-squared (R2) and Root Mean Squared Error (RMSE) to measure predictive accuracy. Additionally, residual analysis, learning curves, and error plots are employed to gain insights into model behaviour.Results from this study offer valuable guidance for selecting the most appropriate regression model for precipitation forecasting in different scenarios. Understanding the strengths and weaknesses of various models can lead to improved accuracy in predicting this critical meteorological variable, ultimately benefiting applications ranging from agriculture to disaster preparedness.
DOI LINK : 10.56726/IRJMETS45024 https://www.doi.org/10.56726/IRJMETS45024