Atmospheric Weather Fluctuation Prediction Using Machine Learning

Abstract

AbstractWeather forecasting predicts atmospheric conditions in the future. Since the manufacturing and livestock businesses depend on precise weather forecasts, it is crucial. Different categories of neural networks are presented in this study in context to efficient weather prediction. The three neural network models for weather prediction investigated in this study are the recurrent neural network (RNN), multilayer perceptron (MLP), and radial basis function (RBF) networks. This paper also went through the steps that were taken to achieve the desired results. Weather is a complex and nonlinear mechanism that can be handled by a neural network. This study uses NumPy, Pandas, Keras, Git, TensorFlow, Matplotlib, Google Cloud Resources, and Anaconda for weather forecasting. The Root Mean Square Error between the predicted and actual values as well as the accuracy of prediction are primarily used to evaluate and compare the outcomes of these models. It was observed that RNN generated the best performance and recorded a minimum RMSE value of 1.432 in a prediction window of 56 days. The accuracy rates produced by RNN, MLP, and RBF were 94.3, 91.5, and 92.9%, respectively. Thus, RNN is considered to be the most efficient at forecasting the weather. As a result, using a machine learning approach along with a recurrent neural network for weather fluctuation prediction provides more relevant and detailed information at a lower cost than using other prediction models.KeywordsNeural networkRecurrent neural network (RNN)Weather forecastingMultilayer perceptron (MLP)Radial basis function (RBF)