Analysis of Different Machine Learning Techniques with PCA in the Diagnosis of Breast Cancer

Abstract

In recent years, different types of cancer cases are common. Increasing cancer cases, A rapidly increasing health for countries and humanity becomes a problem. In addition to being the most common cancer among women today, breast cancer has surpassed lung cancer as the most common cancer type in the world since 2021. Early diagnosis greatly reduces the risk of death in breast cancer, and benign tumors are correctly diagnosed, allows the classification of this field to be a new research topic. New developments in the field of Medicine and Technology Machine learning, classification algorithms and computerized diagnosis are used in the correct classification of tumors. increased its use. These systems are extremely important in terms of being an assistant to the expert opinion. In this study, in the Wisconsin Breast Cancer dataset, it is aimed to accelerate the diagnosis of the disease and to reduce the tumors, different machine learning to minimize treatment processes by providing accurate classification techniques were used. In this study, we reduced our dataset to 171 data using Principal Component Analysis (PCA) to accelerate disease diagnosis on the Wisconsin Breast Cancer dataset and 2 different classification processes were performed using 5 different machine learning. The success rate of each algorithm was compared, and it was revealed that Logistic Regression was the most successful method with an accuracy rate of 98.8% after PCA.