Data Mining Algorithms for Business Decision-Making

This paper presents a comparative analysis of the classification algorithms Logistic regression, Naive Bayes, and K Nearest Neighbors and the results obtained for a business decision-making problem. The performance of the classifiers is evaluated in the terms of accuracy, root mean squared error, and Kappa statistical coefficients. Performance analysis of prediction accuracy demonstrates the effectiveness of the K-Nearest Neighbor algorithm compared with Logistic regression and Naive Bayes, a higher percentage of classified cases, the smallest root mean squared error and the highest degree of prediction according to the true class. The running results in Weka showed that K-Nearest Neighbor has achieved the most correctly classified instances with the least predictive error, compared with Logistic regression and Naive Bayes.