Forecasting Multivariate Sensor data streams

Period: 01.01.2017
- 01.01.2020

Supervisors:

Professor 

Dr. sc. ing.
Dmitry Pavlyuk
Engineering Faculty
Dmitry Pavlyuk

Academic degree: doctoral degree in Economics (2005) and Engineering (2015).

Current position at TSI: professor (since 2019), pro-dean of the Engineering Faculty (since 2021), researcher and head of Data Analytics and Artificial Intelligence research cluster at Transport and Telecommunication Institute, TSI (since 2020).

Experience: academic career from an assistant position at Saratov State University, Russia, in 2002 to an associate professor at TSI. Chair of TSI Mathematical Methods and Modelling Department (2014-2020).

Academic experience: Nineteen years of teaching experience in higher education, development and presenting study courses: Probability Theory and Mathematical Statistics; Econometrics; Operation Research; Optimisation Theory; Discrete Mathematics; Introduction to Stochastic Processes; Data Analysis and Business Forecasting. Member of TSI final attestation commissions on MSc in Computer Science and Management of Information Systems. Dr Dmitry Pavlyuk has supervised 15 successfully promoted MSc and 10 BSc theses. Teaching mobility to Higher School of Transport, Bulgaria (2013), Vilnius Gediminas Technical University, Lithuania (2017).

Research experience: Author of more than 50 publications, including 22 journal articles in Sensors (ISSN: 1424-8220), European Transport Research Review (ISSN: 1867-0717), Algorithms (ISSN: 1999-4893), Transport and Telecommunication (ISSN: 1407-6179), Research in Transportation Economics (ISSN: 0739-8859), Transport (ISSN: 1648-4142) and 12 book chapters in Springer’s Lecture Notes in Networks and Advances in Intelligent Systems and Computing book series. Postdoctoral researcher at TSI (2017-2020) for spatiotemporal urban traffic modelling; presenter at more than 30 international conferences in Latvia, Austria, Cyprus, France, Greece, Germany, Poland, Russia, Spain. The reviewer at many top scientific journals, including Transportation Research, Parts A, C, E, Transport Reviews, IEEE Access, Transportmetrica B. Guest editor at Information (ISSN: 2078-2489); member of programme committees of Reliability and Statistics in Transportation and Communication (RelStat-2017-2020), Latvia, and Computer Modelling in Decision Making (CMDM 2017-2020), Russia. Member of PhD promotion committee, expert of the Latvian Council of Science (since 2017) in Civil Engineering.

Participation in research projects: Dr Pavlyuk has successfully completed the postdoc research project 1.1.1.2/VIAA/1/16/112 “Spatiotemporal urban traffic modelling using big data” (2017-2020); researcher in “Enhancing excellence and innovation capacity in sustainable transport interchanges” (Horizon 2020 Nr. 692426, 2016-2018), “Learning with ICT use” (Erasmus+, 2014-2017) and several commercial research projects.

Research Interests: Research interests include: spatial and spatiotemporal statistical modelling, multivariate time series analysis, econometric models and statistical estimators, stochastic frontier models and machine learning in applied domains of banking, airport industry, urban road traffic, public transport, venture markets, and environmental management.

Project Type: Methodological

Main Challenge:

Sensors are everywhere – from city-wide inductor loops and camera for monitoring urban traffic flows to wearable sensors and widgets. Sensors generate a data flow of enormous volume and velocity, and processing of these data streams is a core part of this research direction.

Funding: ERDF, 1.1.1.2/VIAA/1/16/112

DA&AI tools:

  • Multivariate (spatiotemporal) time series forecasting
  • Artificial neural networks
  • Random Forests
  • Feature Selection

Publications:

  • Pavlyuk, D., 2020. Random Forest Variable Selection for Sparse Vector Autoregressive Models, in: Valenzuela, O., Rojas, F., Pomares, H., Rojas, I. (Eds.), Theory and Applications of Time Series Analysis. Selected Contributions from ITISE 2019., Contributions to Statistics.
  • Pavlyuk, D., 2020. Make It Flat: Multidimensional Scaling of Citywide Traffic Data, in: Kabashkin, I., Yatskiv, I., Prentkovskis, O. (Eds.), Reliability and Statistics in Transportation and Communication, Lecture Notes in Networks and Systems. Springer International Publishing, Cham, pp. 80–89. https://doi.org/10.1007/978-3-030-44610-9_9
  • Pavlyuk, D., 2019. Random Forest-controlled Sparsity of High-Dimensional Vector Autoregressive Models, in: Valenzuela, O., Rojas, F., Pomares, H., Rojas, I. (Eds.), ITISE 2019. International Conference on Time Series and Forecasting. Proceedings of Papers. Presented at the International Conference on Time Series and Forecasting (ITISE 2019), Godel Impresiones Digitales S.L., Granada, Spain, pp. 343–354.

Student Theses:

  • Mertens, E., 2019, Study of Spatiotemporal Feature Selection Methods for Urban Traffic Flow Forecasting. MSc Theses.
  • Farkhshatov, I., 2018, Short-term forecasting of Baltic states’ GDP growth. MSc Theses.

Products:

contact us