Unified Approach to Trajectory Data Mining and Multi-Aspect Trajectory Analysis with MAT-Tools Framework

• Tarlis Tortelli Portela Instituto Federal de Educação, Ciência e Tecnologia do Paraná (IFPR) http://orcid.org/0000-0001-7405-3330
• Vanessa Lago Machado Instituto Federal Sul-Rio-Grandense (IFSul) https://orcid.org/0000-0002-0406-5907
• Chiara Renso ISTI-CNR https://orcid.org/0000-0002-1763-2966

DOI: https://doi.org/10.5753/sbbd_estendido.2024.242862

Resumo

Multiple-aspect trajectory (MAT) data mining requires sophisticated tools to handle the complexity and volume of complex data. This paper introduces MAT-Tools, a comprehensive Python framework for MAT data mining. The framework consists of five main packages: mat-data, which supports data preprocessing and synthetic dataset generation; mat-model, offering model classes tailored for MAT data; mat-similarity, providing methods for similarity measurement; mat-view, visualization tools for MAT data, experimental preparation, and results exploration on a web interface; and mat-classification and mat-clustering, which includes advanced classification and clustering algorithms. Each package addresses specific challenges in MAT data analysis, from preprocessing to modeling and classification. MAT-Tools facilitates efficient and accurate trajectory data analysis, making it invaluable for diverse tasks since exploratory data analysis, anomaly detection, and predictive modeling applications. This framework's integration and extensibility empower researchers and practitioners to gain deeper insights and achieve more reliable results in trajectory data mining.

Palavras-chave: data mining, python, trajectory analysis, similarity, classification, clustering

Referências

de Freitas, N. A., da Silva, T. C., de Macêdo, J. F., Junior, L. M., and Cordeiro, M. (2021). Using deep learning for trajectory classification. In Proceedings of the 13th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART, pages 664–671. INSTICC, SciTePress.

Ferrero, C. A., Petry, L. M., Alvares, L. O., Leite da Silva, C., Zalewski, W., and Bogorny, V. (2020). MasterMovelets: discovering heterogeneous movelets for multiple aspect trajectory classification. Data Mining and Knowledge Discovery, 34(3):652–680.

Leite da Silva, C., May Petry, L., and Bogorny, V. (2019). A survey and comparison of trajectory classification methods. In 2019 8th Brazilian Conference on Intelligent Systems (BRACIS), pages 788–793.

Lettich, F., Pugliese, C., Renso, C., and Pinelli, F. (2023). Semantic enrichment of mobility data: A comprehensive methodology and the MAT-BUILDER system. IEEE Access, 11:90857–90875.

Mello, R. d. S., Bogorny, V., Alvares, L. O., Santana, L. H. Z., Ferrero, C. A., Frozza, A. A., Schreiner, G. A., and Renso, C. (2019). MASTER: A multiple aspect view on trajectories. Transactions in GIS, 23(4):805–822.

Petry, L. M., Ferrero, C. A., Alvares, L. O., Renso, C., and Bogorny, V. (2019). Towards semantic-aware multiple-aspect trajectory similarity measuring. Transactions in GIS, 23(5):960–975.

Petry, L. M., Leite da Silva, C., Esuli, A., Renso, C., and Bogorny, V. (2020). MARC: a robust method for multiple-aspect trajectory classification via space, time, and semantic embeddings. International Journal of Geographical Information Science, 34(7):1428–1450.

Portela, T. T., Bogorny, V., Bernasconi, A., and Renso, C. (2022a). Automatise: Multiple aspect trajectory data mining tool library. In 2022 23rd IEEE International Conference on Mobile Data Management (MDM), pages 282–285.

Portela, T. T., Carvalho, J. T., and Bogorny, V. (2022b). Hiper-movelets: high-performance movelet extraction for trajectory classification. International Journal of Geographical Information Science, 36(5):1012–1036.

Portela, T. T., Machado, V. L., Carvalho, J. T., Bogorny, V., Bernasconi, A., and Renso, C. (2024). Ultramovelets: Efficient movelet extraction for multiple aspect trajectory classification. In Database and Expert Systems Applications (DEXA), pages 79–94, Cham. Springer Nature Switzerland.

Vicenzi, F., Petry, L. M., Silva, C. L. D., Alvares, L. O., and Bogorny, V. (2020). Exploring frequency-based approaches for efficient trajectory classification. Proceedings of the ACM Symposium on Applied Computing, pages 624–631.

Viera-López, G., Morgado-Vega, J., Reyes, A., Altshuler, E., Almeida-Cruz, Y., and Manganini, G. (2023). pactus: A python framework for trajectory classification. Journal of Open Source Software, 8(89):5738.