Institutional Repository of Vilnius University
Mašininio mokymosi metodų taikymas vertinant drabužių tinkamumą /
Abstract
dc:descriptionIn this bachelor’s thesis, the focus is on applying machine learning methods to improve the fit of women's and men's clothing in e-commerce, aiming to reduce the return rate of products. The goal is to develop models that can effectively assess clothing suitability for users with different body dimensions. The analysis is based on classification and clustering methods. The study examines the suitability of women's and men's clothing using machine learning techniques. The main objective is to optimize the clothing recommendation system based on users' body measurements. The study employs K-means clustering, UMAP visualization, t-SNE analysis, and various classification algorithms such as logistic regression, decision tree, and XGBoost. The results showed that in the women's group, logistic regression achieved 60,5\% accuracy, decision tree 74,4\%, and XGBoost 72,1\%. In the men's group, all three classifiers achieved equal accuracy of 73,3\%. Although accuracy was equal in the men's group, and the decision tree classifier showed better accuracy in the women's group, XGBoost was chosen as the best classifier as it also evaluated mean error and standard deviation using cross-validation with the best parameter grid. The accuracy of the men's model with cross-validation was 0,712 with a standard deviation of 0,019, and for women, it was 0,715 with a standard deviation of 0,085. Significant features were also identified for both men's and women's groups. Applying and improving these models can significantly enhance user satisfaction and optimize the fitting system, reducing logistical costs and return rates. The thesis provides suggestions for model improvement and practical application to ensure the best possible size match in real-world trading situations.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Gelžinytė, Simona,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD210643435&prefLang=en_US