An Intelligent Multi-Task Framework for Used Vehicle Valuation Marketability Classification and Budget-Aware Recommendation
Engineering Systems and Intelligent Technologies (ESIT) · 2026
DOI: 10.66279/svgz0r91 ↗Kurz gefasst
Fachartikel in der Zeitschrift Engineering Systems and Intelligent Technologies (ESIT), 2026. Verfasst von Rasha Y. Youssef und 3 weiteren Autor:innen. Die Arbeit nennt 38 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
An intelligent decision support framework for used car evaluation, pricing, and recommendation is presented, integrating acceptability assessment, price estimation, marketability classification, and budget aware recommendation. The Car Evaluation benchmark is used for vehicle evaluation. After removing nineteen invalid marker rows from the 1,729 row source file, 1,710 valid, label consistent instances remain, with no duplicates, and preprocessing including SMOTE balancing is performed within training folds only. Ten classifiers are evaluated using stratified ten fold cross validation with grid search optimization and a soft voting ensemble, reaching a cross validated accuracy of 0.9985, with the leading tree based and boosting models statistically indistinguishable on the test set. A synthetic Egyptian used car dataset of 6,000 listings, generated from documented depreciation and pricing rules, supports the market oriented tasks. CatBoost gives the strongest price estimation performance, with an MAE of 44,768 EGP, a MAPE of 6.0 percent, and an ????2 of 0.986, significantly outperforming Random Forest. The marketability classifier reaches 0.605 accuracy across three demand classes, matching a logistic regression baseline given the deliberately noisy demand label. The recommendation engine, evaluated on 200 simulated buyer queries against a transparent ground truth, achieves a Precision at 5 of 0.88 and an NDCG at 10 of 0.82, measuring recovery of the generator’s hidden signals rather than validated real world quality, since the ground truth shares the engine’s utility formula. The results demonstrate feasibility for integrating predictive analytics, market assessment, and recommendation into a single framework, with real listing validation left for future work.
Zitieren
APA
Youssef, R. Y., Eldeen, D. A., Elsayed, S. M., & Emadeldein, A. (2026). An Intelligent Multi-Task Framework for Used Vehicle Valuation Marketability Classification and Budget-Aware Recommendation. Engineering Systems and Intelligent Technologies (ESIT). https://doi.org/10.66279/svgz0r91
Vancouver
Youssef RY, Eldeen DA, Elsayed SM, Emadeldein A. An Intelligent Multi-Task Framework for Used Vehicle Valuation Marketability Classification and Budget-Aware Recommendation. Engineering Systems and Intelligent Technologies (ESIT). 2026. doi:10.66279/svgz0r91
BibTeX
@article{Youssef2026,
author = {Youssef, Rasha Y. and Eldeen, Donia Alaa and Elsayed, Shaimaa Mohamed and Emadeldein, Ahmed},
title = {An Intelligent Multi-Task Framework for Used Vehicle Valuation Marketability Classification and Budget-Aware Recommendation},
journal = {Engineering Systems and Intelligent Technologies (ESIT)},
year = {2026},
doi = {10.66279/svgz0r91}
}
EUVIMED