Performance enhancement of machine learning algorithms for general medical dataset classification and predicting future trends of fuel consumption in indian transportation

Abstract

Classification is a data mining technique, used to predict class newlinemembership for data instances. Classifier performance depends greatly on the newlinecharacteristics of the data to be classified. Real world data may consist of newlineredundant and conflicting instances, irrelevant and redundant attributes. Thus the newlinedata need to be preprocessed prior to classification. Feature selection is the newlineprocess of selecting a subset of features in the training set and using only this newlinesubset as features in data classification. It makes training and applying a newlineclassifier more efficient by reducing the size of the feature space. It often newlineincreases classification accuracy by eliminating irrelevant and redundant newlinefeatures. newlineThe first component of this research work focuses on enhancing the newlineperformance of k-Nearest Neighbor algorithm for effective data classification. newlineThe k-NN algorithm is amongst the simplest of all machine learning algorithms newlinein which an object is classified by a majority vote of its neighbors, with the newlineobject being assigned to the class that is most common amongst its k nearest newlineneighbours. newline