Experimental Analysis of Precipitation Forecasting Using Machine Learning and Distributed Machine Learning Approach

Weather forecasting plays a significant role in the field of agriculture, energy industry, forestry and provides stable habitat for a wide variety of species. The prediction methods play a vital role in accurate precipitation forecasts. This chapter aims to evaluate the accuracy of rainfall prediction in India through machine learning (ML) techniques like K-Nearest Neighbors (KNN), Decision Tree (DT), Random Forest (RF), and Ada-Boost (AB). The high spatial resolution data like rainfall, maximum and minimum temperature data were collected from all over India from the Indian Meteorological Department (IMD) (6.5 N° & 66.5° E to 38.5° N & 100.5° E). The datasets for validation/training and testing consisted of daily rainfall maximum and minimum temperature from different locations within India from 2018 to 2022. The dataset is represented in the form of a grid, which is arranged in the degree of 1.0*1.0. We have evaluated and compared the machine learning algorithms and distributed machine learning algorithms using Dask-ML. The Dask-ML enables distributed machine learning with data parallelism by efficiently processing extensive datasets across a cluster of machines, scaling workflows and handles large volumes of data. Distributed machine learning algorithms have faster execution.