Significance of Attribute Selection In The Classification of Chronic Renal Disease

• Published in: 2019 Second International Conference on Advanced Computational and Communication Paradigms (ICACCP) pp. 1 - 6
• Main Authors: Banerjee, Anonnya, Noor, Alaa, Siddiqua, Nasrin, Uddin, Mohammed Nazim
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.02.2019
• Subjects: Attribute Selection; CFS; Chronic Kidney Disease; classification; Classification algorithms; Data mining; Diseases; GreedyStepWise; Kidney; Logistics; Prediction algorithms; Predictive models
• DOI: 10.1109/ICACCP.2019.8882937

The management and diagnosis of CKD or Renal Failure is often complicated. Maximum risk associated with lack of knowledge, poor health management, and indifferent lifestyle can be considered as primary reasons for increasing cause of this disease. Basically, Chronic Renal Disease is intensifying the damage of kidneys over a duration. Decreased or damaged kidney is often determined using GFR, which is also known as golemerular filtration rate. However, it can be cured or prevented if diagnosed and identified at an early stage or else dialysis and transplantation will be needed for further mishaps. Identifying CKD is can be very complex, proper medical test are needed. Several attributes from the medical test can reveal many useful information about Chronic Kidney Disease. To successfully use these features the significance of these features should be analyzed with greater details in order to provide the required treatment or cure. In this paper, classification models were suggested using different classifying algorithms, Permutation Feature Importance, Correlation-based Feature Subset attribute evaluator and GreedyStepwise search methods are used to classify and identify CKD and non-CKD patients. From the obtained result, it was observed that classifiers performed better with reduced dataset attributes as compare to the original.