Breast Cancer Identification and Classification Using Machine Learning Techniques

Abstract

The most prevalent disease among women  and a major contributor to the rising mortality rate among  women is breast cancer. There is a high need for automatic  diagnostic systems for early identification of breast cancer  since manual breast cancer diagnosis takes a lot of time and  there are few systems available. Deep learning and  machine learning approaches play a vital role in  developing such technologies. We have employed machine  learning classification algorithms to distinguish between  benign and malignant tumours. These approaches allow  the computer to learn from previous data and predict the  category of fresh input. Breast cancer is the major cancer  in women (43.3 instances per 100,000 women) but the one  with the highest mortality rate (14.3 incidents per 100.000 women). For survival, early diagnosis is essential. The  issue may be successfully identified, forecasted, and  evaluated using machine learning techniques. The eight  machine learning methods we compared in this study were  the Gaussian Naive Bayes (GNB), kNearest The  experiment includes a confusion matrix, 5-fold merge,  Neighbours (K-NN), Svm Classifiers (SVM), Variational  Forest (RF), AdaBoost, Multilayer Perceptron (GB), and  datasets from Breast Cancer Badgers. The results of the  tests showed that ANN had the better spec. Efficiency  (99.28 percent), F1-score (99.99 percent), recall (96.75  percent), accuracy (99.19 percent), and AUC were attained  via XGBoost (99,61 percent ). Our findings demonstrated  that, in the Breast Cancer Wisconsin dataset, ANN is the  most successful approach for predicting tumor.