A WRAPPER BASED FEATURE SELECTION APPROACH USING BEES ALGORITHM FOR EXTREME RAINFALL PREDICTION VIA WEATHER PATTERN RECOGNITION THROUGH SVM CLASSIFIER

Rainfall prediction is a major problem in metrological department and it is closely associated with the economy and life of human. Accuracy of rainfall prediction is a very important for countries like india, because Indian economy is mainly dependent up on the agriculture. Most of the statistical methods are unsuccessful due to dynamic nature of atmosphere. In this paper, SVM classifier is used for classification, before SVM classifier, bees algorithm is used for feature selection. Keeping all the features and instances in the training set is not a good approach for better classification because all the features are not contributing more information during classification, removing those features will not affect any classification accuracy. The experimental result shows that the proposed method provides better detection rate, false positive rate, accuracy rate and less training time of classifier what we obtained with entire dataset.