Performance Comparison of Classification Algorithm in Datamining Techniques using Chronic Kidney Dataset

The problem of chronic kidney disease is getting worsened day by day. It is also known as chronic renal disease and is a life threatening disease; it has various symptoms such as high blood pressure, anemia, rashes, muscle pain, conjuctivitus, etc. So, in order to tackle this problem it has to be detected at earliest stages possible and given suitable treatment before it get worsened. We have used 24 symptoms of chronic kidney disease in this paper which help us to accurately detect this disease with the help of eight classification algorithms i.e. SGD, Random subspace, SMO, JRIP rules, Hoeffding tree, NaiveBayes, Locally weighted learning, oneR in data mining tool WEKA. We conclude the results by introducing the medical datasets to all three algorithms separately with the help of knowledge flow interface of WEKA data mining tool, the parameters which are used to compare the results of these three different algorithms are mean absolute error, kappa statistics and total number of instances studied either correctly or incorrectly. The main aim of this paper is present a clear view of the chronic kidney disease, its symptoms and the criteria to detect it at earliest stages possible which will help the mankind to get safe from this life threatening disease.