Performance Analysis of Averaged Perceptron Machine Learning Classifier for Breast Cancer Detection

Abstract

Breast cancer is the primary cause of women's death due to cancer; if detected in the early stage, it is a curable disease. Machine learning classification techniques are helpful in breast cancer detection. The research aims to investigate the averaged-perceptron machine-learning classifier performance on the Wisconsin original breast cancer dataset (WBC); the work has focused on two points; first, does the averaged-perceptron classifier has the quality to gain a higher accuracy than the other classifiers? Second, does it help to reduce false-negative or false-positive breast cancer predictions? The averaged-perceptron model recorded an accuracy score of 0.984 with zero false-negative predictions. The investigation has also signified the effect of threshold on false-negative or false-positive prediction. Applying the averaged-perceptron classifier in a computer-aided-diagnosis system can improve breast cancer recognition accuracy with zero false-positive or false-negative forecasts.