A study on the evaluation of different regressors in Weather Prediction

Abstract

Due to its applicability in the actual world with problems such as meteorology, agricultural studies, and so on, weather prediction has become a very significant research topic. Weather forecasting is the process of forecasting the state of the atmosphere using several climatic characteristics. Present atmospheric condition is gathered and analyzed for weather forecasts. Meteorologists and academics have found accurate weather prediction to be a difficult endeavor. Weather data is critical in agriculture, tourism, airports, mining, and power generating are just a few examples. The rapid growth in the generation of meteorological data as well as the progress of climate observing technologies such as satellite meteorological observation led to the Big Data era. In this paper, we are analyzing weather data sets using different regressors. A detailed study on regression using machine learning regression models, namely, Linear Regressor, Polynomial Regressor, Decision Tree Regressor, Random Forest Regressor, Linear BayesianRidge Regressor, Linear Ridge Regressor, Linear Lasso Regressor, Linear ElasticNet Regressor, Support vector Regressor, and Artificial Neural Network(ANN) regressor are presented in this paper. Further, the performance of regressor models was measured through the error rate in the prediction with MSE, RMSE, MAE, and R-squared measures. Experimental results reveal that Random Forest regressors and Decision Tree regressors give better performance compared to other machine learning regressors. Regression using an Artificial Neural Network gives the best results compared to the machine learning approach in terms of prediction rate and execution time. This study helps in forecasting future weather conditions that farmer who grows crops by monitoring weather patterns and also arranging cricket, football matches, and open-ground events. In coastal areas, they anticipate tsunamis and natural calamities.