Comparative analysis of weather prediction using ensemble learning models and neural network

Abstract

Understanding nature and projecting its future behavior has always been deemed to be a major breakthrough for survival and moving our civilization forward. In the age of Machine learning, data mining, data analysis, this particular process has received a dramatic improvement. This paper converged our focus into several significant machine learning algorithms to analyse their contributions to predicting weather patterns. Having access to a 20-year data of a single station has given us a centralized view to compare and determine which algorithm is better suited for weather forecasting. We have worked on AdaBoosting, XGboosting, Stacking KNN, Stacking Neural Network etc. to validate their performances and differentiate their scores on Confusion matrix measurements. We have found some astonishing achievements by these algorithms, mainly Stacking Neural Network, and distinguished their performances.