An Empirical model of Data Analysis and Techniques for Breast Cancer Detection

Abstract

—Breast cancer is a common and deadly disease 
that affects millions of women worldwide. Early detection 
is crucial for successful treatment, and machine learning 
techniques can help improve the accuracy of breast cancer 
diagnosis. In this work, the breast cancer dataset obtained 
from Kaggleis analysed using multiple decomposition 
techniques and classification algorithms, with a particular
focus on model optimization.
PCA, Sparse PCA, Fast ICA, and NMF were applied for 
dimensionality reduction, and Naive Bayes, Logistic 
Regression, Kernel SVM,SVM,KNN and Random Forest 
for classification. A combination of hyper parameter tuning 
and the model performance is optimized using crossvalidation.