Classification of Malignant and Benign tissue with Logistic Regression

Abstract

Detection of breast cancer is the major phase in Cancer Diagnosis. So, classifiers with higher accuracy are always superior. A classifier already carrying high accuracy and then leading it to higher accuracy offers very less chance to a patient to be wrongly classified. This book investigates the use of a modified and improved version of the hypothesis used in the logistic regression. Both gradient descent and advanced optimization techniques are used for the minimization of the cost function. A weighting factor β was assigned in the hypothesis which is a sigmoid function. The dependency of this weighting factor to the number of features, the size of the dataset and the type of optimization technique used were observed. The accuracy was improved significantly by appropriately choosing the value of β, which, is a function of both the number of features and the type of optimization techniques used. The obtained results using the weights were promising, resulting in a significant increase in accuracy, sensitivity, and specificity.