Inquisition of The Support Vector Machine Classifier in Association with Hyper-parameter Tuning: A Disease Prognostication Model

Although medical data classification is a challenging task, it allures the research community so that it can help to take precise precautions in restraining future diseases. To perpetrate the operations, many Machine Learning (ML) algorithms have been extensively used where data pre-processing steps play a monumental role to comprehend the algorithms with better performances. This exploration shows a outstretched analysis of the Support Vector Machine (SVM) algorithm to suggest a model for prognosis of Breast Cancer(BCC), Cleveland Heart Disease (CHD) and Parkinson Disease (PD). To accomplish this experiment, primarily the pre-processing techniques such as categorical data encoding, resampling, feature selection and scaling etc. are performed followed by evaluating the performances of SVM with 10-fold cross-validation technique after using hyper-parameter tuning. Though many experiments are demonstrated in the literature, there is still scope for upliftment. Our proposed model analyses the datasets and improves the performance over the existing classification models which is proved by comparing with several previous studies as well as five other ML classifiers' performances conducted by us by passing through the same steps as the SVM. Hence, mostly better performances are manifested concerning other existing studies where the accuracy obtained by our proposed model are delivered the best of 94.49% in BCC dataset, 94.54% in CHD dataset and 94.91% in PD dataset.