Development of Machine Learning models using WEKA for Atmospheric Data

Abstract

There are number of Atmospheric variables such as rainfall, temperature, wind speed, humidity, visibility, Wind gust, Precipitation, etc. For learning covariance of machines by principles, practical and probabilistic approaches are made using Gaussian process. In this paper by taking visibility, time with date, temperature as independent or responding variables and wind speed as dependent or response variable, we fit Gaussian process model. K star is an instance based classifier that classifies the data. RBF network is used for data and is similar structure of Gaussian process but it uses clustering method with weight parameters. Additive regression classifies the variables by using Decision Stump. Decision Tree Regression improves the model by removing the decisions of the tree that are not important in classification. We fit different Waikato Environment for Knowledge Analysis (WEKA) models for atmospheric data and which model is the best based on RMSE values.