Applying Data Mining and Classification Techniques for Detecting Diabetes Mellitus

Purpose - this research has focused on developing an ensemble system using data-mining methods basedon three classification methods namely, weighted k-nearest neighbor, simple decision tree and logisticregression algorithms to detect diabetes mellitus of the human. According to the World HealthOrganization, the seventh major cause of human death in 2030 will be diabetes and of course is a verysevere disease which, if not treated thoroughly and on time, can lead to critical difficulties, includingdeath. Consequently, diabetes is one of the main priorities in medical science researches, which usuallyleads to the production of lots of information. The role of data mining methods in diabetes research iscritical which considered as one of the optimum procedures of extracting knowledge from a large amountof diabetes-related data. Methodology - the proposed ensemble method uses votes given by the each ofthe algorithms to produce the final result. This voting mechanism considers each estimation of theclassifiers as an input to the ensemble system and then computes the statistical mode for its output to getthe majority vote. Results - apparently, these classifiers give the accuracy of 77.00%, 77.30%, 79.30%and 80.60% for Decision Tree, Weighted K-nearest neighbor, Logistic Regression and the ensemblemethod respectively. Conclusion - the results of the proposed method illustrate an acceptableimprovement of accuracy compared to other methods. Consequently, it supports the idea that Hybridmethods in data classification are more effective in comparison with the simple classification methodswhich use classifiers separately.

کلیدواژه ها:

Data mining ، decision tree ، logistic regression ، weighted K-nearest neighbor ، diabetes mellitus ، classification ، cross validation