Indian Journal of Science and Technology

Abstract

Objectives: The model built in this paper is used to classify the rainfall datasets in identifying districts of more rainfall and of lesser rainfall in the state of Andhra Pradesh. Methods: In this paper support vector machine, random forest, Knearest neighbor and decision tree classification methods have been used to classify rainfall data sets which is divided into training set and test set for classification and later validation of the obtained results. Findings: Based on various statistical parameters like sensitivity, prevalence, detection rate, specificity, and detection prevalence it has been concluded that support vector machine classification methods is better than any other classification method used in the research. Rainfall data sets are used to initially build the classification model and the results are tested against the test set. Using the confusion matrix thus obtained the mentioned statistical parameters are obtained to establish the supremacy of support vector machine classification method. Applications: Examples of satellite imagery has become ever more significant in numerous application domains such as ecology monitoring and alternative discovery. Rainfall classification is the application used herein.

Keywords: Classification, Data Mining, Classifier, Support Vector Machines, SVM