Manuscript Rainfall Prediction using Multilayer Perceptron (MLP)

In this study, we undertook an analysis focused on forecasting hourly rainfall. This analysis framed the problem as binary classification, with rainfall events classified into two categories: “rainy” (the positive class) and “non-rainy” (the negative class). We leveraged independent climatic variables from the current hour to predict rainfall conditions for the following hour. Our data source was the CST weather station, providing records of 8 hourly weather parameters. To make predictions, we harnessed a commonly used machine learning model, the Multilayer Perceptron (MLP), resulting in an accuracy rate of 79%.