| Abstrakt: |
Nowadays, the major health problem seen globally is Chronic Kidney Disease (CKD). It has high morbidity and mortality rate, and it induces other diseases. Unlike other diseases, this doesn’t show any symptoms in the early stage, so patients often fail to notice the disease. Early detection of CKD enables patients to receive timely treatment to enhance the progression of this disease. The major challengence faced by physician is mostly patients fail to miss some medical history. Machine learning models can effectively help physicians achieve this goal due to their fast and accurate recognition performance. The datasets which contain medicial history of the patients contain irrelevant features that may negatively affect the disease diagnosis efficiently. Inorder to avoid these problems the datasets are imputed using various imputation methods. Then the imputed data are fed to classifiers to observe the performance of the model. The commonly used machine learning classifier methods are Logistic Regression (LOG), Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Naive Bayes (NBs) and Feed Forward Neural Network (FFNN). In this study K Nearest Neighbor (KNN) imputation was used to fill in the missing values, which selects several complete samples with the most similar measurements to process the missing data for each incomplete sample. Then the performance is observed by feeding the data in various classifiers and the results are compared against each type. The CKD dataset was obtained from the University of California Irvine (UCI) machine learning repository, which has a large number of missing values. The classifier and existing methods are evaluated using the metrics like precision, recall, F1-score, sensitivity, specificity, and accuracy. [ABSTRACT FROM AUTHOR] |
