Improved Weather Prediction Mechanism with Mode Based Approach

This Weather prediction is a complex application that needs multiple phases of processing. The processing application built through the proposed mechanism using mode based pre-processing along with clustering mechanism. Mode is the highest value from the attribute. By replacing mode value with the missing value, correlation may become negative. This negative correlated attributes have to be eliminated from the dataset. The mechanism used first of all load the dataset. After loading the dataset, pre-processing is applied. The outliers and missing values are tackled with the help of mode based approach. Missing values are represented with "NA" and can be handled by replacing these missing values with the calculated mode. The outliers are extreme values present within the dataset. These extreme values will be replaced with the average values to increase the classification accuracy of overall operation. Labelling information is applied with correlation and clustering mechanism. Labeling information allows to accurately predict the weather. The classification is done using with clustering and multiple classes has been predicted. The result of classification accuracy is in the range of 80% that is improved by 4% using proposed approach. The demonstration of the work is done on the dataset derived from benchmarked website.