Advancing Breast Cancer Diagnosis: Ensemble Machine Learning Approach with Preprocessing and Feature Engineering

Considering the rise in instances over the past 10 years, breast cancer remains a serious health danger, particularly for women. Different approaches have been put forth to identify and categorize breast tumours, with an emphasis on differentiating between cases that are benign and those that are malignant. A thorough approach is used in this work to create a reliable system for classifying breast cancer cases. A number of machine learning models, including LDA (Linear Discriminant Analysis), Cat Boost Classifier, LGBM Classifier, XG Boost, Random Forest Classifier, Extra Trees Classifier, and Ridge Classifier, are used in the study, which makes use of the WBCD (Wisconsin Breast Cancer Diagnostic) dataset. Preprocessing techniques like correlation analysis are used in the workflow to extract pertinent features and create new features based on correlated values. To address class imbalances in the dataset, SMOTE analysis is also carried out. After the models are trained and their accuracy assessed. The REXCRLL-E model is built to take advantage of each model’s unique strengths to enhance classification performance. By combining various classifiers into a single ensemble system for more precise and dependable tumour classification. Experimental results show that proposed model performance reaches 98.83%.