Volume 18 Issue 1 (2025)

A modern approach to evaluating car prices using Machine Learning algorithms

Nickolas-Alessandro Afrem, Paul Stefan Popescu

University of Craiova
Al. I. Cuza Street, 13, Craiova, Romania

Abstract. The global automotive industry, especially the secondary market for used vehicles, operates in a dynamic and complex economic environment. This environment has significant gaps in information. This paper describes the development and use of a complete web application designed for predicting used car prices and offering intelligent vehicle recommendations. The application was built using ASP.NET Core Blazor Server and features a Machine Learning pipeline powered by ML.NET and a FastTree Regression algorithm trained on a dataset of approximately 290,000 vehicles listed on the German market. By integrating strong backend ML processing with external APIs like the RapidAPI VIN Decoder and Google Gemini AI, the platform reduces data entry challenges and improves output clarity. The predictive model achieves an R-Squared score of 0.9174 with a mean absolute error of €2,798, supported by a Retrieval-Augmented Generation (RAG) system that grounds AI suggestions in real-world conditions. This paper outlines the comprehensive development process, covering everything from data preprocessing and algorithm setup to building a user-friendly, real-time interface for quantitative deal analysis.

Keywords: machine learning, FastTree Regression, retrieval-augmented generation

Cite this paper as:
Afrem, N.-A., Popescu, P. S. A modern approach to evaluating car prices using Machine Learning algorithms. International Journal of User-System Interaction 18(1), 1-23, 2025. DOI: 10.37789/ijusi.2025.18.1.1

Agent-Based Operationalisation of Health Behaviour Change Interventions

Ciprian Amaritei, Dragoș Silion, Adrian Iftene

“Alexandru Ioan Cuza” University
Bulevardul Carol I, Nr.11, 700506, Iaşi, România

Abstract. Designing Health Behaviour Change applications requires translating behavioural theory into effective digital interventions. This paper presents a proof-of-concept operationalisation of the VITAL framework through a multi-agent architecture designed to support product designers in creating behaviour change interventions. The architecture is organised into behavioural design, UI strategy, and UI component generation modules. Specialised agents support the selection of process motivators and Behaviour Change Techniques, the translation of behavioural strategies into interface concepts, and the generation of prototype components. The approach is illustrated through a hydration case study, showing how intermediate behavioural decisions can remain linked to resulting interface artefacts. The findings demonstrate the technical feasibility of the proposed workflow and its ability to preserve traceability between behavioural rationale and design outcomes. Further evaluation is needed to assess its effectiveness, consistency, and usefulness in practice.

Keywords: health behaviour change, multi-agent systems, artificial intelligence, digital health, behaviour change techniques

Cite this paper as:
Amaritei, C., Silion, D., Iftene, A. Agent-Based Operationalisation of Health Behaviour Change Interventions. International Journal of User-System Interaction 18(1), 24-37, 2025. DOI: 10.37789/ijusi.2025.18.1.2