Machine Learning-Based Prediction of Consumer Purchasing Behaviour in the Organic Cosmetic Products

This study presents an approach to predict consumer purchasing behaviour for organic cosmetic products such as cleansers, blushes, lipsticks, and serums using machine learning models. The proposed study employs a machine learning-based predictive framework using a dataset representing key product features such as formulation type, product category, and brand associations. The models were trained to predict consumer ratings based on purchasing intention. Model performance was evaluated using regression metrics such as R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> Score, Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The linear models performed best, showing extremely high coefficients of determination (greater than 0.999) with almost negligible error rates: RMSE = 1.95, MAE = 1.55, and MAPE = 4.5%. Applied ensemble methods performed differently, with XGBoost and Gradient Boosting yielding R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> values of 0.89. The study also computed and visualized feature importance, applied feature selection techniques, and explored feature extraction methods. The proposed methodology offers a scalable solution for market research and strategic planning in consumer-focused industries.