Abstract:

Chronic kidney disease is a global health problem with increased prevalence and occurrence, poor outcomes, and increased costs. The consequences of chronic kidney disease not only include kidney failure but also reduced kidney function complications. Current research indicates that early identification and diagnosis will eliminate or postpone some of those adverse effects. The use of technology is promising in this field, so we are using machine learning to help in early recognition of CKD .we are using algorithms like Support Vector Machine (SVM), K-Nearest Neighbour (KNN), Decision Tree and Random Forest comparing between their results to achieve the maximum accuracy and having two different datasets "Pima Indians Diabetes dataset" and "Chronic Kidney Disease dataset," with varying numbers of the attribute (8 for the first dataset and 25 for the second dataset) and of course different attribute values. We are also training the machine on various percentages of training set size (70% and 80%) and test set size (30% and 20%). And different K values (1 and 3) for the KNN. In the end we found out that by training the algorithm on 80% of the data in the "Chronic Kidney Disease" dataset and by using the Random Forest method, we could achieve an accuracy of 100%.