Research on Breast Cancer Classification and Prediction Model Based on Principal Component Analysis and Machine Learning

Abstract

In recent years, breast cancer has become one of the biggest threats to women's health, accounting for the majority of cancer deaths among women. Because early treatment of breast cancer has a great effect on the recovery of breast cancer, the diagnosis of breast cancer is particularly important. Machine learning, as the most popular method, is also used for model construction in this field. This study is based on data from breast tumors, which contain 10 morphological features of breast tumor nucleus. In this study, homogenization, standardization and feature selection were used to process the data and KNN algorithm was used to construct the classification prediction model, with principal component analysis (PCA) used to optimize the model. Finally, the original 30 variables were reduced to 3 variables and the model parameters were adjusted in order to achieve the best model with the accuracy of 98.54%. The final model achieves the highest operating efficiency and accuracy. In this study, through the visualization of PCA and the comparison of different models, the classification effect of the final model can be the best. This model can be applied to the clinical diagnosis of breast cancer patients, which is helpful to the early treatment efficiency and greatly reduce the mortality of breast cancer patients