Published Paper Details:

Supervised machine learning algorithms play a critical role in solving classification problems across science and engineering. This study presents a comparative evaluation of four widely used supervised classification models--Logistic Regression, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Decision Tree--using the Iris and Wine benchmark datasets. An experimental design was adopted with an 80/20 stratified train-test split, standard normalization, and label encoding to ensure fair comparison across models. Performance was assessed using accuracy, precision, recall, and F1-score to capture both correctness and robustness of predictions. Results show that SVM consistently achieves the highest accuracy (0.98 for Iris and 0.88 for Wine), followed closely by Logistic Regression, while KNN and Decision Tree generally exhibit lower performance, reflecting sensitivity to feature scaling and over fitting, respectively. These findings confirm that classifier effectiveness strongly depends on dataset characteristics, including separability and feature distributions. The study underscores the necessity of empirical benchmarking when selecting classification algorithms for real-world problems rather than relying on default or heuristic choices. Future research should extend this comparison to ensemble and deep-learning-based classifiers and incorporate larger, higher-dimensional engineering datasets to enhance generalizability