Comparative Analysis of Multiple Machine Learning Algorithms in Heart Disease Classification Prediction

This paper aims to evaluate the effectiveness of various machine learning models in handling classification tasks by comparing and analyzing their performance on two different datasets. The study employs multiple machine learning models, including Random Forest, Multi-Layer Perceptron, K-Nearest Neighbors, XGBoost, Support Vector Machines, AdaBoost, Logistic Regression, Gaussian Naive Bayes, Linear Discriminant Analysis, Decision Trees, and Gradient Boosting. For each model, metrics such as accuracy, precision, F1 score, log loss, Hamming loss, and Jaccard similarity coefficient are calculated on both datasets D1 and D2. Additionally, the performance of the models is further analyzed through Precision-Recall (PR) curves and Area Under the Curve (AUC) curves. On dataset D1, XGBoost, AdaBoost, Logistic Regression, and Gradient Boosting models exhibit outstanding performance in accuracy. The Multi-Layer Perceptron achieves the best precision and Jaccard similarity coefficient, while the Gradient Boosting model performs the best in log loss and Hamming loss. For dataset D2, Logistic Regression obtains the highest values in multiple evaluation metrics, demonstrating its superiority in handling this dataset. Ensemble learning methods show good performance on both datasets, while linear models are notable for their efficiency and interpretability. Ensemble learning methods and linear models exhibit excellent performance in handling classification tasks, especially on complex datasets. Different datasets have significant impacts on model performance, and selecting an appropriate model requires comprehensive consideration of dataset characteristics and task requirements. Future research can further explore new evaluation metrics and methods, as well as cross-domain data fusion and model transfer, to enhance the generalization ability and practicality of machine learning models.