Analysis of Breast Cancer Classification using Various Algorithms

Invasive ductal carcinoma has become the most common types of tumors in recent years. The chances of the patient arriving in the case improve if the issue is detected early. The most prevalent classification for breast cancer is binary (benign cancer/malignant cancer), which allows pathologists to come up with a systematic and objective prognosis.The WBCD (Wisconsin Breast Cancer Diagnosis) dataset is used to build the classifier. Before performing the classification, the dataset is preprocessed and explored using various techniques. Four Machine Learning techniques like k-fold cross-validation, pipelining, principle component analysis(PCA), and hyperparameter optimization are compared and analyzed in the project. These techniques are applied to the development of eight classifiers that must distinguish between benign and malignant breast tumors. These techniques were applied on the Adaboost classifier, KNeighbor classifier, DT classifier, RF classifier, SVM classifier, Logistic regression classifier, Gaussian NB classifier, and Gradient boosting classifier. After applying all those algorithms, the SVM classifier has obtained the highest accuracy of 99.1% after applying all those algorithms on the dataset.