A Comparative Study of Different Classification Algorithms on Kidney Disease Prediction

Rajni Garg
2018 International Journal for Research in Applied Science and Engineering Technology  
Medical wellness is very basic need of everyone in today's society. Medical problems are increasing exponentially day by day, so as data is also increased like medical data, healthcare data, patient data, their treatment and source management data. So it becomes difficult to take the right decision at the right time from the large medical dataset. Medical diagnosis is an important but complicated task that should be performed accurately and efficiently and its automation would be very useful. A
more » ... system for automated medical diagnosis would enhance medical care and reduce costs. Data mining play a vital role to discover the hidden pattern of medical diagnosis. In this paper systematic study on "Data mining in Chronic_Kidney_Disease dataset" has been done. In this study five data mining algorithms are executed on Chronic_Kidney_Disease dataset. Quality of these algorithms is measured on the basis of four parameters like correctly classified instance, incorrectly classified instance, execution time, error rate. The experiment is done using 10 fold cross validation method. The study has proven that accuracy of Naïve Byaes classification algorithm is being considered highest i.e. 98.75% and minimum execution time is taken by random tree i.e. 0.005 seconds. Feature reduction technique is used for identifying and removing those at tributes that do not contribute towards classification of the dataset. In this work, chi square attribute selection technique is used to evaluate the worth of an at tribute with respect to the class. Then ranker algorithm is used to arrange these attributes in descending order according to their chi square value and last 5 attributes having lowest ranked value are removed based on assumption that these attribute don't contribute to the classification. The comparison is also made in this study which shows the affect of computation time and accuracy before feature reduction and after feature reduction applied. Our study shows that after feature reduction most of the algorithms improved their accuracy and computation.
doi:10.22214/ijraset.2018.2132 fatcat:rx7yij27azdrfpapnl2ezu4mn4