Using data mining technique to classify medical data set

Abstract EN

Data mining is an iterative and multistage process to extract a useful knowledge.

the real life data mining applications are interesting because they often present a different set of problems for data miners; one of such applications that we have done is on the diabetic patients databases.

medical data mining is the process of extracting hidden patterns from medical data.

in this research, a data mining method using an enhanced ripper classification algorithm has been designed and implemented.

this algorithm is rule-based classifier which is based on rule growing and pruning .it apply on the pima Indians of national institute of diabetes datasets and Mirjana hospital diabetes datasets in order to satisfy acceptable accuracy percentage of classification rate compare with other existing classification algorithms.

the data set is loaded to the system and split it to patient and non-patient based on the patient class, and then the initial rules are applied to obtain the first classification percent.

information gain is applied in each classified step as an indicator for the important rule.

in the final step, rules are growing based on unification method that applied in each step, where the rules with the high rank are merged to obtain a new rule.

the obtained results were small in the first growing process.

accuracy rate is increased during the merging rules process.

in final iteration, we obtained accuracy detection is higher than in the standard existing algorithm.