Turin, Italy – The five-year agreement between Tierra S.p.A. and Politecnico di Torino is entering its third year, keeping on focusing on R&D in Data Analytics. The last projects were presented by Silvia Buccafusco and Andrea Megaro and were coordinated by Francesco Vaccarino, Associate Professor at the Department of Mathematical Sciences of Politecnico. These projects are part of the Micene Project developed in collaboration with the SmartData Center of Politecnico, represented by Luca Vassio, Assistant Professor and Research Fellow.
“The Micene Project deals with the automatic identification of the status (duties) of the vehicle. We realized that this information is very important for our customers and that it is sometimes not set correctly. We wanted to understand how to support customers by making these statuses increasingly accurate and reliable. In Machine Learning, we use clustering to solve this problem. In our case, clusters corresponded to the statuses.”
Lucia Salvatori – Data Analyst at Tierra S.p.A.
Silvia Buccafusco’s project “Intelligent vehicle states definition: IoT, time series and data analytics” focused on pre-processing, actual analyses and validation of the results. The goal was to automatically identify operational usages of vehicles. This has been achieved applying unsupervised algorithms – clustering – in multidimensional time-series, using fine-grained data (up to 50Hz). Tierra provided the data by collecting them from a test vehicle through CAN-bus transmission, enabling a real use case application. As a successful completion of her achievements, Ms. Buccafusco was awarded full marks with honors at her graduation. She recently joined Tierra Applied Research unit as Data Scientist, becoming at the same time a PhD student in mathematics at Politecnico di Torino (thanks to a joint project Tierra – Politecnico di Torino approved by Regione Piemonte).
Andrea Megaro’s project “Vehicle patterns identification through multivariate time series analysis” focused on univariate and multivariate window clustering and in particular on the analyses, experimenting very numerous different solutions to deepen the question, like the Internal Cluster Validity Indices (CVI) to evaluate the goodness of a cluster based on the inter-cluster and intra-cluster distance to choose the optimal clustering setting. Andrea applied also the Dynamic Time Warping (DTW) method, an algorithm allowing alignment between two sequences to lead to a distance measurement between the two aligned sequences. Mr. Megaro will be graduating in November.
“The Micene project is part of our larger effort to develop new innovative approaches and methodologies, through our long-term collaboration with Politecnico di Torino. These results want to be our future products. We always want to push boundaries to develop new research ideas and transform them into products that benefit our customers.”
Riccardo Loti – Head of Research at Tierra S.p.A.
“The Micene project was a very exciting one, both for its aims and the high quality of the people involved. Despite the difficulty of the proposed tasks, the fantastic team involved was able to obtain very interesting results also enlightened by the use of very advanced methodologies. This latter fact constitutes the starting point of the path toward new tasks and ventures aimed at the realization of more efficient and powerful analytics tools for Tierra. I would like here to congratulate with Silvia and Andrea and to thank all Tierra team, as well as the SmartData Center crew and, last but not least, Prof. Luca Cagliero whose highly professional contribution was a key factor for the success of this project.”
Francesco Vaccarino – Associate Professor, Department of Mathematical Sciences of Politecnico di Torino.