Effective Diabetes Prediction: Integrating Ensemble Learning with LIME for Robust Results

In this study, we developed a robust framework for predicting diabetes using a combination of Machine Learning (ML) algorithms, including Random Forest (RF), Support Vector Machine (SVM), Naive Bayes, Decision Tree (DT), Neural Networks, and K-means clustering. We utilized a comprehensive dataset encompassing medical and demographic features. Our methodology included meticulous data preprocessing, extensive exploratory data analysis (EDA), and the creation of interaction features to capture complex relationships. We applied Recursive Feature Elimination (RFE) to select significant features and employed ensemble learning techniques using a Voting Classifier to enhance model performance. Our Neural Network model demonstrated superior predictive capability, achieving an accuracy of 97.21% and an ROC-AUC score of 0.9769. To ensure model interpretability, we implemented Local Interpretable Model-agnostic Explanations (LIME), which provided clear insights into individual feature contributions. This approach not only ensures high accuracy in diabetes prediction but also enhances transparency and trustworthiness, making the results valuable for clinical decision-making.