Accuracy Study Based on Different Preprocessing Machine Learning Approaches

With the ever-increasing data in the world, artificial intelligence (AI) is predicted to be an essential tool for converting huge data into meaningful data. The goals of machine learning (ML) are to identify complicated patterns in multi-dimensional data and utilize uncovered patterns to classify processes or predict data. Prediction accuracy is the accustomed measure of success in ML applications. In this paper, the effect of using a machine learning approach on accuracy value is studied. We compared three feature selection algorithms, namely IndFeat, MRMR, and ANOVA algorithms. The SVM classifier is applied to the selected features in order to compare the three algorithms and examine the effect of data reduction on the accuracy value. The simulation results showed that the ANOVA algorithm is better than the MRMR and IndFeat algorithms.