Problem 1

Personalized Forecasting of Total Trip Duration Based on Driver Classification from Instantaneous Velocity Data

Brief Description of the Challenge:

Accurately predicting trip duration is critical for improving traffic and telecommunication management. The goal of this challenge is to evaluate whether the addition of the driver's type can improve the accuracy of forecasts for the total trip duration. To achieve this, we have a dataset based on real observations containing instantaneous velocity, total trip duration, and driver types (categorized as lower, middle, and speediest) from various cars traveling a specific route from Guimaraes to Braga.

This challenge can be divided into the following steps:

1) Driver Classification: The first challenge is to classify the type of driver using the minimum number of instantaneous velocity values during the initial phase of the trip, necessary to ensure accurate classification.

2) Personalized Prediction: Once the driver type is determined, the next step is to make a personalized prediction of the total trip duration based on this classification. This approach will be compared with other prediction methods to evaluate its performance.

Mathematical background:

Participants need to have basic knowledge of a programming language, data analysis and predictive modeling.

Coordinators:

Flora Ferreira, Paulo Araújo e Susana Faria, Universidade do Minho, Portugal