RecomiGlowTech: Cosmetic Recommendation System

Abstract

"The research's objective is to create a cosmetic recommendation system that accurately suggests products to users based on their preferred input. Customers can choose the best product for their skin type and make quick decisions due to this. This is very convenient and saves a great deal of time. In this study, content-based filtering is employed because it offers a greater degree of personalization by meeting particular needs. Additionally, there isn't a problem with cold start, which is a big problem for recommendation systems and does not impact content-based systems. Additionally, it lessens the impact of popularity bias because content-based recommendations prioritize user preferences over popularity, which reduces the distortion caused by popularity bias.. Various ML models like Random Forest, Logistic Regression, SVM, Naive Bayes and Decision Trees were tested and Random Forest was chosen as it had the highest accuracy. Three data science metrics were used to evaluate the system's performance: accuracy, recall, and F1-score. The results showed that the recommendation engine achieved a high level of recall and precision, indicating that the system was successful in suggesting appropriate cosmetic products to users. The website's user interface was also evaluated through usability testing, and the findings showed that the design was easy to understand and operate."