EUVIMEDEUVIMED entdecken
Crossref

MODELING ARTIFICIAL INTELLIGENCE METHODS FOR PREDICTING THE ADDITIONAL VALUE OF CONSUMER TRANSACTIONS

Nazarii Dykyi, Yulian Salo

Computer Design Systems. Theory and Practice · 2026 · 8 (1) · 39-48

DOI: 10.23939/cds2026.01.039 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer Design Systems. Theory and Practice, online erschienen am 30.04.2026. Verfasst von Nazarii Dykyi und 1 weiteren Autor:innen. Die Arbeit nennt 13 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper investigates the application of machine learning methods for predicting the additional value of customer transactions aimed at improving the efficiency of commercial departments. The relevance of the study is driven by the rapid growth of customer data volumes and the need to transition from intuition-based decision-making to model-driven approaches. Particular attention is paid to upselling and cross-selling tasks, where classification errors can lead to direct financial consequences. For the experimental study, a synthetic dataset was constructed to simulate typical customer behavior in a B2C environment. The dataset includes transaction frequency, average order value, customer interaction history, and aggregated indicators of purchasing activity. This approach enables control over data structure, noise levels, and class balance, ensuring a fair and reproducible comparison of the evaluated models. Three predictive models were implemented: a baseline logistic regression model, a cost-sensitive logistic regression model, and a random forest ensemble model. Model performance was assessed using a set of classification metrics, including overall classification accuracy, precision, recall, F1-score, and the area under the ROC curve, as well as an expected revenue metric. The results indicate that the random forest model achieved the strongest discriminative performance (ROC AUC of approximately 0.72) and demonstrated the highest potential for practical deployment in upselling and cross-selling scenarios. At the same time, logistic regression provided stable and interpretable results, making it suitable for environments with strict explainability requirements, while the cost-sensitive approach did not yield significant improvements without additional fine-tuning. Overall, the findings confirm the effectiveness of ensemble methods for predicting additional transaction value and emphasize the importance of integrating traditional classification metrics with economically oriented evaluation criteria.

Zitieren

APA

Dykyi, N., & Salo, Y. (2026). MODELING ARTIFICIAL INTELLIGENCE METHODS FOR PREDICTING THE ADDITIONAL VALUE OF CONSUMER TRANSACTIONS. Computer Design Systems. Theory and Practice, 8(1), 39-48. https://doi.org/10.23939/cds2026.01.039

Vancouver

Dykyi N, Salo Y. MODELING ARTIFICIAL INTELLIGENCE METHODS FOR PREDICTING THE ADDITIONAL VALUE OF CONSUMER TRANSACTIONS. Computer Design Systems. Theory and Practice. 2026;8(1):39-48. doi:10.23939/cds2026.01.039

BibTeX
@article{Dykyi2026,
  author = {Dykyi, Nazarii and Salo, Yulian},
  title = {MODELING ARTIFICIAL INTELLIGENCE METHODS FOR PREDICTING THE ADDITIONAL VALUE OF CONSUMER TRANSACTIONS},
  journal = {Computer Design Systems. Theory and Practice},
  year = {2026},
  volume = {8},
  number = {1},
  pages = {39--48},
  doi = {10.23939/cds2026.01.039}
}
Kostenlos starten →