Main Article Content
The condition known as chronic kidney disease (CKD) is widespread worldwide and accounts for a significant number of deaths. With 1.2 million deaths annually, chronic kidney disease (CKD) ranks 11th on the global death toll. According to the kidney Foundation of Bangladesh, approximately 40,000 CKD patients experience kidney failure each year and several thousand die in the early stages of their lives as a result of CKD. It is challenging to use machine learning in predictive analytics for healthcare to assist physicians in selecting the most effective treatments for saving lives. The majority of the collaborative research conducted by scientists on chronic kidney diseases relied solely on statistical models, resulting in numerous development gaps for machine-learning models. In this project, we evaluated two pre-processing scenarios, combined significant characteristics of the F1 scores, and discussed the current methods and suggested improved technology based on the correlation. In addition, we provided clinical information-based machine learning techniques for anticipating chronic renal disease. The Random Forest Classifier (RFC) and the Logistic Regression (LR) are two master teaching strategies that are investigated. The components are derived from the UCI chronic kidney disease dataset, and the best regression model for the prediction is chosen by comparing the results of these models. From these two preprocessing scenarios, choosing important features and replacing missing values with mean values for each column was the most logical choice because it lets you train with more data without dropping. However, correlation produced the best results in both instances, with an accuracy of 98%. As a result, the system can be used to predict early stage CKD at a low cost, which will be helpful for developing and underdeveloped nations.