Style Mate: A Comparative Analysis of Outfit Compatibility Predictor in Fashion Technology Employing Machine Learning Models

The e-commerce industry is quickly evolving because of the AI implementation. The AI intelligence is required to predict an intelligent outfit suggestion to improve the customer experience and it also enhances the customer personalization. The approaches for Machine learning are very effective to dominate the fashion recommendation problems. This study evaluates performances of 4 classical machine learning technique - XGBoost, Random Forest (RF), K-Nearest Neighbors (KNN), Naïve Bayes - in analyzing the outfit compatibility between clothing items like pants and shirts.. A collected resources was pre-processed and feature engineered with visual indicators such as color histogram, texture and shape embeddings. The models are trained and validated while performances are evaluated through Accuracy, Precision, Recall, ROC- AUC curve. Comparative analysis shows that XGBoost is highly suitable for fashion recommended systems, continuously outperforming Random Forest and KNN in terms of classification accuracy and generalization.