AI-Driven Evaluation of Electric Motorcycles A Cloud-Native Framework Integrating TOPSIS Strategy, Gradient Boosting, and Large Language Models on Azure Databricks
DOI:
https://doi.org/10.15662/IJEETR.2025.0706012Keywords:
Electric motorcycles, TOPSIS, gradient boosting, XGBoost, LightGBM, large language models, Azure Databricks, cloud-native, multi-criteria decision making, explainable AI, feature importance, EV evaluationAbstract
This paper proposes a cloud-native framework for systematic evaluation and ranking of electric motorcycles (e-motorcycles) by combining multi-criteria decision making (MCDM) via TOPSIS, supervised learning via gradient boosting, and natural language understanding provided by large language models (LLMs), all orchestrated on Azure Databricks. The framework addresses two common industry needs: (1) integrate heterogeneous data sources — telemetry, financial/total-cost-of-ownership (TCO) estimates, user reviews, regulatory/charging infrastructure indicators — and (2) produce transparent, explainable, and operationally deployable rankings for procurement, consumer guidance, and fleet planning. Data ingestion and pre-processing occur in Azure Databricks using Apache Spark for scalable ETL; feature engineering includes domain-derived features (battery energy density, motor efficiency, range-per-charge, charging time, TCO components) and textual features extracted from maintenance logs and user reviews using an on-cluster LLM-based embedding and summarization pipeline. A gradient boosting model (XGBoost/LightGBM) predicts target operational performance metrics (e.g., real-world range, mean-time-between-failures, TCO deviation) and produces feature importance scores that feed into the TOPSIS weight calibration process. TOPSIS performs multi-criteria ranking using hybrid weights (domain + model-derived importances) to produce ranked candidate lists; the framework supports sensitivity analysis, scenario-based ranking (e.g., urban vs. highway fleets), and counterfactual queries answered via an LLM assistant component. We evaluate the framework on a mixed dataset (synthetic + industry-sourced telemetry and reviews) and report improvements in rank stability, predictive accuracy (RMSE reductions vs. baseline), and user-aligned ranking fidelity compared with standard MCDM-only and ML-only baselines. We discuss deployment considerations, cost-performance trade-offs on Azure Databricks, and avenues for real-time decisioning. The proposed architecture aims to improve fleet acquisition decisions, consumer transparency, and continual evaluation pipelines for e-mobility stakeholders.
References
1. Hwang, C. L., & Yoon, K. (1981). Multiple Attribute Decision Making: Methods and Applications. Springer.
2. Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. Annals of Statistics, 29(5), 1189–1232.
3. Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 785–794). ACM.
4. Raju, L. H. V., & Sugumar, R. (2025, June). Improving jaccard and dice during cancerous skin segmentation with UNet approach compared to SegNet. In AIP Conference Proceedings (Vol. 3267, No. 1, p. 020271). AIP Publishing LLC.
5. Poornima, G., & Anand, L. (2025). Medical image fusion model using CT and MRI images based on dual scale weighted fusion based residual attention network with encoder-decoder architecture. Biomedical Signal Processing and Control, 108, 107932.
6. Kiran, A., Rubini, P., & Kumar, S. S. (2025). Comprehensive review of privacy, utility and fairness offered by synthetic data. IEEE Access.
7. Amuda, K. K., Kumbum, P. K., Adari, V. K., Chunduru, V. K., & Gonepally, S. (2024). Evaluation of crime rate prediction using machine learning and deep learning for GRA method. Data Analytics and Artificial Intelligence, 4 (3).
8. Kakulavaram, S. R. (2023). Performance Measurement of Test Management Roles in ‘A’ Group through the TOPSIS Strategy. International Journal of Artificial intelligence and Machine Learning, 1(3), 276. https://doi.org/10.55124/jaim.v1i3.276
9. Adari, V. K. (2024). APIs and open banking: Driving interoperability in the financial sector. International Journal of Research in Computer Applications and Information Technology (IJRCAIT), 7(2), 2015–2024.
10. Kandula, N. Innovative Fabrication of Advanced Robots Using The Waspas Method A New Era In Robotics Engineering. IJRMLT 2025, 1, 1–13. [Google Scholar] [CrossRef]
11. Archana, R., & Anand, L. (2025). Residual u-net with Self-Attention based deep convolutional adaptive capsule network for liver cancer segmentation and classification. Biomedical Signal Processing and Control, 105, 107665.
12. Konda, S. K. (2025). LEVERAGING CLOUD-BASED ANALYTICS FOR PERFORMANCE OPTIMIZATION IN INTELLIGENT BUILDING SYSTEMS. International Journal of Research Publications in Engineering, Technology and Management (IJRPETM), 8(1), 11770-11785.
13. Kiran, A., & Kumar, S. A methodology and an empirical analysis to determine the most suitable synthetic data generator. IEEE Access 12, 12209–12228 (2024).
14. Bussu, V. R. R. Leveraging AI with Databricks and Azure Data Lake Storage. https://pdfs.semanticscholar.org/cef5/9d7415eb5be2bcb1602b81c6c1acbd7e5cdf.pdf
15. Dhanorkar, T., Kotapati, V. B. R., & Sethuraman, S. (2025). Programmable Banking Rails:: The Next Evolution of Open Banking APIs. Journal of Knowledge Learning and Science Technology ISSN: 2959-6386 (online), 4(1), 121-129.
16. Balaji, P. C., & Sugumar, R. (2025, June). Multi-level thresholding of RGB images using Mayfly algorithm comparison with Bat algorithm. In AIP Conference Proceedings (Vol. 3267, No. 1, p. 020180). AIP Publishing LLC.
17. Kesavan, E. (2023). Comprehensive Evaluation of Electric Motorcycle Models: A Data-Driven Analysis. Intelligence, 2, 1. https://d1wqtxts1xzle7.cloudfront.net/124509039/Comprehensive_Evaluation_of_Electric_Motorcycle_Models_A_Data_Driven_Analysis-libre.pdf?1757229025=&response-content-disposition=inline%3B+filename%3DComprehensive_Evaluation_of_Electric_Mot.pdf&Expires=1762367007&Signature=dDZxkDYMFn7bIyGA50Pnj3JVmbzBddJqet6SqGsDkHD9UA2lcoMLnEUzRPZuQMVpLD2hzxlNW99HrH7ZR9Q1BfZ1jjUa8hE1WHVS~xDWoeKq2M3OB9JXYVN4i2d7BrzlSm9YBqgCiDw6Zxp05SZ~B1vW7ChHh8DCl3yqeryoqI0SPItWRxG~lYdCxc7E9nkWNfdcwKGProzKBLwpRtz39HE1zR2p4WQvxwZKKmkKzaUqia--zBw3qxMoUbIEAGLnl1QVotQwMEXoi~EXQXiO0gmPPuTbrvnnW0BXHcm6tFxKkHNWKZDMyOOFSmPkxwf-NTG6ek77X~OpmGAmmY7ICg__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA
18. Asaduzzaman M, Dhakal K, Rahman MM, Rahman MM, Nahar S. Optimizing Indoor Positioning in Large Environments: AI. Journal of Information Systems Engineering and Management [Internet]. 2025 May 19 [cited 2025 Aug 25];10(48s):254–60. Available from: https://jisemjournal.com/index.php/journal/article/view/9500
19. Phani Santhosh Sivaraju, 2025. "Phased Enterprise Data Migration Strategies: Achieving Regulatory Compliance in Wholesale Banking Cloud Transformations," Journal of Artificial Intelligence General science (JAIGS) ISSN:3006- 4023, Open Knowledge, vol. 8(1), pages 291-306.
20. Gorle, S., Christadoss, J., & Sethuraman, S. (2025). Explainable Gradient-Boosting Classifier for SQL Query Performance Anomaly Detection. American Journal of Cognitive Computing and AI Systems, 9, 54-87..
21. Bishop, C. M. (2006). Pattern Recognition and Machine Learning. Springer.
22. Molnar, C. (2020). Interpretable Machine Learning: A Guide for Making Black Box Models Explainable (book; available online). — discusses SHAP and model explainability techniques used to connect feature importances to decision processes.
