Synergizing Intelligence: A Comparative Evaluation of Machine Learning and Data Mining Techniques for Optimized Computational Analytics

Abstract

The growing importance of data-driven decision making in healthcare, and also other domains like intelligent analytics, has motivated the need for models that are able to achieve high predictive performance while still being interpretable. However, prior works often focused on data mining that contribute transparent rule-based reasoning or machine learning model with good predictive power instead of successfully integrating the two. To fill this gap, a hybrid model of Computational Analytics (HCAM) was proposed and evaluated that unified datamining–method-based feature structuring with supervised machine learning’s method-based prediction in an analytical pipeline. We initially performed K-Means clustering, Apriori association rule mining and Decision Tree–based feature selection to identify meaningful patterns along with filtering-in of less redundant features. The pruned feature matrix was then applied to fit Support Vector Machine, Random Forest, XGBoost and Artificial Neural Network classifiers for ultimate prediction. We implemented the framework in Python and tested it on the UCI Heart Disease dataset with 303 samples after standard preprocessing (including normalization and categorical encoding). The performance was evaluated with 10-fold cross-validation, and statistical significance was tested by one-way ANOVA followed by Tukey’s HSD. Results Experimental results demonstrated that the hybrid model performed better than all other models, achieving 92.8% accuracy, 92.7% for F1-score and AUC = 0.97. These gains were shown to be supported statistically at p < 0.05. The results showed that the hybrid modelling improved interpretability in addition to predictive validity for structured analytical problems.