Evaluating Label Encoding and Preprocessing Techniques for Breast Cancer Prediction Using Machine Learning Algorithms
Mukesh Kumar, Vivek Bhardwaj
Abstract
Breast Cancer (BC) is a major health concern in the world, accounting for a disproportionately high number of new cases each year among females. Its frequency as a health issue has increased significantly in recent years. Early detection of BC is the most straightforward method of coping with the diagnosis. In 2020, approximately 2.3 million women worldwide were diagnosed with breast cancer, resulting in around 685,000 deaths. Therefore, early detection of BC is crucial for effective treatment and improved survival rates. The results and assessments of many Machines Learning (ML) models for detecting BC survivability are presented in this manuscript using the BC dataset. Although the dataset is relatively small, it provides valuable insights. The data was evaluated with many ML techniques and different predictive models were built with the valuable results. Several ML techniques were applied to build predictive models, including Gaussian Naïve Bayes (GNB), k-Nearest Neighbors (k-NN), Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM). Data scaling and encoding techniques, including StandardScaler and MinMaxScaler, are employed to enhance the accuracy of these machine learning models. Additionally, preprocessing steps, such as Numerical Variable Correlation, Categorical Variables Analysis, Continuous Variables Analysis, Bivariate Analysis, Balancing Classes (oversampling function) are applied to enhance the model’s performance. The results show that out of all the ML techniques tested, the k-NN method gives the most accurate predictions, which is close to 94.00%.