Feature Reduction Set for the Prediction of Renal Disease Using Ensemble Methods and Optimal Hyperplane Algorithms

The rising incidence of long-term renal ailment is a serious issue for worldwide public health. The rate of mortality is usually associated with the illness, particularly in poorer nations. Since there are no visible early-stage signs, kidney disease frequently goes undiagnosed. In the meantime, preventing the disease from progressing requires early detection and prompt clinical care. To help clinicians discover kidney disease early, models such as Machine Learning can offer efficient and affordable computer-aided diagnostic approaches. This method could reduce the duration and expense of renal disease screening, hence a small subset of clinical test characteristics would be required for the diagnosis. Diagnosing the disease at its initial stage has become a challenging task for physicians. ML algorithms such as Random Forest (RF), Artificial Neural Network (ANN), Support Vector Machine (SVM), Naïve Bayes (NB), and Decision Tree (DT) played a major role in predicting the disease at the initial stage. The UCI Repository data consisting of 24 attributes was taken to do the prediction using the above-mentioned algorithms. Here the Random Forest is considered to acquire the condensed set of critical 12 features for forecasting renal illness, and the prediction accuracy is improved with a reduced number of attributes. Boosting algorithms like Gradient boost and Adaptive boost are compared with Optimum Hyperplane (SVM) to effectively estimate the accuracy in predicting the disease; an accuracy of 98.33% is achieved with gradient boost, thereby saving life and allowing them to lead a normal life for the rest of the period.