Collaborative Filtering Based Approach for Personalized Next Basket Recommendation on Fashion Products

Pardosi, Juan Carlos Tepanus (2023) Collaborative Filtering Based Approach for Personalized Next Basket Recommendation on Fashion Products. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05111942000017-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (4MB) | Request a copy

Abstract

Preferences that are dynamic and always change with the times are characteristic of fashion enthusiasts. This means all businesses in the fashion sector have to adapt to this very dynamic change. To solve this problem, it is necessary to have the next basket recommendations for fashion products. A collaborative filtering-based approach is one of the methods that can be used to provide recommendations for the next basket. The data set used in this study comes from transaction data for fashion products, which has been divided based on four seasons, such as fall, winter, spring, and summer. Alternating Least Squares (ALS) and Non-negative Matrix Factorization (NMF) are two collaborative filtering algorithms used in this study to provide the next basket recommendations for each customer. By analyzing a customer's purchase history and finding hidden patterns, this method uses a factorization matrix to obtain important information regarding consumer trends and preferences.
In this research, the implementation of collaborative filtering with the ALS algorithm and the NMF algorithm was carried out. As the basis for evaluating the performance of the recommendation system that has been made, the Mean Average Precision (MAP) is used in the top 12 items. The model that has the highest MAP@12 values in each season is the model used in the development of the recommending system. After doing the experiments using the ALS algorithm, the highest value was obtained for MAP@12 items in each season: fall with a value of 0.002797159, winter with a value of 0.001518113, spring with a value of 0.002352652, and summer with a value of 0.001754891. Besides that, the experiments were also carried out using the NMF algorithm, and the highest score was obtained on MAP@12 items in each season: fall with a value of 0.000670914, winter with a value of 0.001063630, spring with a value of 0.001790324, and summer with a value of 0.001080172. Based on the results of this research, it was found that the ALS algorithm is more effective and accurate in providing recommendations for the next basket when compared to the NMF algorithm in a collaborative filtering approach.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Next Basket Recommendations, Collaborative Filtering, Recommendation System, Alternating Least Squares, Non-negative Matrix Factorization, Rekomendasi Keranjang Belanja Berikutnya, Penyaringan Kolaboratif, Sistem Rekomendasi, Alternating Least Squares, Non-negative Matrix Factorization.
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Juan Carlos Tepanus Pardosi
Date Deposited:	28 Jul 2023 15:57
Last Modified:	28 Jul 2023 15:57
URI:	http://repository.its.ac.id/id/eprint/99675

Actions (login required)

View Item