การปรับความไม่สมดุลของข้อมูลด้วยการจำแนก 5 วิธี

Abstract

We compared the imbalanced data of four methods; i.e. over sampling, synthetic minority over sampling technique, under sampling, and hybrid methods, using five classification methods; i.e. k-nearest neighbor, artificial neural network, support vector machine, rule-based, and stochastic gradient descent. Metrics were accuracy, sensitivity, specificity, mean square error and mean absolute error. The data sets were chemotherapy for stage B/C colon cancer, monoclonal gammopathy and treatment of migraine headaches. Each of these data sets was divided into three proportions in the ratio of 70:20:10 using the data part 1. Training data are used to create a model 70 percentages; the data part 2. Validation data are used to evaluate an error a model 20 percentages, and the data part 3, testing data are used to test a model 10 percentages using the random seed 10, 20, 30, 40, and 50 by WEKA program. When we compared the chemotherapy for stage B/C colon cancer data set, the monoclonal gammopathy data sets, and the treatment of migraine headaches data sets, the best method was the ruled-based in imbalanced data adapting the synthetic minority over sampling technique.