Cancer Prediciton with different Machine Learning Techniques with different sizes of Running and Testing Sets

Abstract

The medical field is new entrant in the machine learning based prediction. There are plenty of machine learning techniques in the category of supervised learning and unsupervised learning which can be  applied on the complete dataset for the purpose of prediction. In the current research, three algorithms in the category of supervised machine learning technique have been applied on the cancer dataset. There are various parameters for the patient test result is being considered for the purpose of prediction. The whole dataset is tested with different sizes of training and testing sets. The results for K-NN, SVM and logistic regression have been applied for training and testing sets. The optimum technique in the form of logistic regression is evaluated with the highest accuracy of prediction by 98.16% at training and testing partition ratio of 30:70.