EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Hybrid of Deep Auto-Encoder and Feature Embedding Model for an E-Commerce Recommender System

Justin Murithi Ireri, Edna Chebet Too, Moses Mahugu Muraya

East African Journal of Information Technology · 06.05.2026 · 9 (1) · 220-233

DOI: 10.37284/eajit.9.1.4936 ↗

Kurz gefasst

Fachartikel in der Zeitschrift East African Journal of Information Technology, online erschienen am 06.05.2026. Verfasst von Justin Murithi Ireri und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study has developed a hybrid model using a deep neural network and feature embedding to increase accuracy to solve the cold start problem on a product-based e-commerce recommender system. A recommender system aims to predict users’ interests and recommend product items that are quite likely to be interesting to them. Recommender systems have been used in many different platforms, such as online movies, online shopping, music stores, and many more, to provide personalised recommendations to users. Models based on traditional machine learning methods, such as collaborative and content-based algorithms, have faced low prediction accuracy, sparsity issues, and the cold start problem. A cold start problem arises when a recommender system fails to predict the appropriate item for a new user, often because it lacks sufficient information to learn and make a personalised recommendation. The model has incorporated a feature-embedding layer into a deep auto-encoder neural network, which automatically extracts user time and location data points. The data point increases user information, hence having a rich profile, increasing item prediction accuracy up to 96% compared to other base models. The model has been trained and tested using the Keras TensorFlow running on the Google Colab platform. We used the MovieLens 100k dataset from Kaggle, and the model was evaluated using Root Mean Squared Error (RMSE) and Mean Squared Error (MSE). This research contributes to providing an accurate personalised shopping experience, increasing sales, and efficient marketing.

Zitieren

APA

Ireri, J. M., Too, E. C., & Muraya, M. M. (2026). A Hybrid of Deep Auto-Encoder and Feature Embedding Model for an E-Commerce Recommender System. East African Journal of Information Technology, 9(1), 220-233. https://doi.org/10.37284/eajit.9.1.4936

Vancouver

Ireri JM, Too EC, Muraya MM. A Hybrid of Deep Auto-Encoder and Feature Embedding Model for an E-Commerce Recommender System. East African Journal of Information Technology. 2026;9(1):220-233. doi:10.37284/eajit.9.1.4936

BibTeX
@article{Ireri2026,
  author = {Ireri, Justin Murithi and Too, Edna Chebet and Muraya, Moses Mahugu},
  title = {A Hybrid of Deep Auto-Encoder and Feature Embedding Model for an E-Commerce Recommender System},
  journal = {East African Journal of Information Technology},
  year = {2026},
  volume = {9},
  number = {1},
  pages = {220--233},
  doi = {10.37284/eajit.9.1.4936}
}
Kostenlos starten →