Abstract
The objective of the paper is to analyze the rainfall data using the logistic regression model. More precisely, we are interested in finding out the most influential variables (effects) behind the rainfall in Bangladesh. In doing this, we fit seasonal logistic regression models (SLRMs). Odds ratio of the predictor variables of SLRMs are computed to detect the most influential variables; relative efficiency of the SLRMs are also calculated to find out the most efficient season for rainfall, and several factors are extracted for each season. Consequently, the Kharif logistic regression model has the highest relative efficiency comparing to Rabi and Pre-kharif logistic regression models, and the estimates of odds ratio for Rabi and Kharif seasons are greater than that of the Pre-Kharif season. We have found the most influential variables for rainfall in Rabi and Kharif seasons.
