Optimized Preprocessing using Time Variant Particle Swarm Optimization (TVPSO) and Deep Learning on Rainfall Data

Abstract

In the recent past, rainfall prediction has played a significant role in the meteorology department. Changes in rainfall might affect the world's manufacturing and service sectors. Rainfall prediction is a substantial progression in giving input data for weather information and hydrological development applications. In machine learning, accurate and efficient rainfall predictionis used to support strategy for watershed management. The prediction of rain is a problematic occurrence and endures to be a challenging task. This paper implements a novel algorithm for preprocessing and optimization using historical weather from a collection of various weather parameters. The Moving Average-Probabilistic Regression Filtering (MV-PRF) method eliminates unwanted samples with less amplitude from the database. The Time Variant Particle Swarm Optimization (TVPSO) model optimizes the preprocessing rainfall data. Then this optimized data is used for the different classification processes. The preprocessing methods emphasize the recent rainfall data of the time series to improve the rainfall forecast using classification methods. Machine Learning (ML) technique classifies the weather parameters to predict rainfall daily or monthly. These experimental results show that the proposed methods are efficient and accurate for rainfall analysis.