Impact on Data Mining Classification of Principal Component Analysis in Hypertension Estimation
In this paper, 150 people who were 30 years old or older and who did not use any medicines consistently from the possible information on hypertension; gender, age, lipid profile, triglyceride, body mass index, uric acid and cigarette use data were collected and a hypertension database was created. Of these, 65 are healthy and the remaining ones are hypertensive diseases. First, Naive Bayes, Multilayer Sensor Network, Decision Table and C4.5 classification algorithms were implemented on this database. Then, by implementing Principal Component Analysis, the size of hypertension database was reduced and the same classification algorithms were applied again and the results were compared. All of the algorithms except from the Naive Bayes classifier showed that the Principal Component Analysis improved classification accuracy.
Keywords - Hypertension Estimation, MLP, Naïve Bayes, C4.5, Principal Component Analysis