A Study on the Application of Machine Learning Algorithms Using R

Abstract

Machine learning is preferred to human interpretations during the analysis of vast scientific datasets, since the data processing time is reduced with and increased accuracy of results. Gene classification is very important in scientific analysis of bio assay datasets especially for effective disease identification and drug discovery. The important task in gene classification process is the construction of decision tree. Real time datasets are used for this analysis and the results are in terms of 35 known siRNA gene expression factors of 64752 substances. Three types of machine learning algorithms are used for analyzing the bio assay dataset which are CTree algorithm, Rpart algorithm and K-Means clustering algorithm. The performance analysis shows better accuracy, precision and F-measure rates of Rpart algorithm. From the result analysis, it is inferred that the performance tree based decision making algorithms like Rpart and CTree is far better than K-Means clustering algorithm in the diagnosis of biological activities in the protein coded genes. It is also proved that the tree based algorithms are less sensitive and less specific which leads to a development of high impact decision trees even for complex datasets like AID_651811. The analysis is done in R programming tool and the mutation factors are predicted and classified.