Welcome to the repository for our paper, "A Data-Driven Framework for Driving Cycle Generation and Analysis." Here, we present a novel methodology combining Markov chain modeling, Monte Carlo simulation, and Dynamic Time Warping to generate realistic driving cycles, specifically tailored for the unique traffic conditions of Igdir, Turkey. Explore our code and datasets to see how we accurately model driving patterns to aid in sustainable mobility solutions.
- Generation of realistic driving cycles using advanced techniques.
- Incorporation of Markov chain modeling, Monte Carlo simulations, and Dynamic Time Warping (DTW).
- Accurate representation of the complexity and stochastic nature of real-world driving behavior.
Pseudo-code for the proposed algorithm.
The dataset provided in this repository consists of 19 driving cycles collected from the city of Igdir, Turkey. The driving cycles were recorded using an On-Board Diagnostic II (OBD-II) device installed in a 5-door vehicle. The data was acquired onboard using an Android device.
The driving cycles cover a route of approximately 11,150 meters, including two main arteries (D080 and E99 international trade routes) connecting Turkey to Asian countries. The dataset includes driving cycles during both peak and off-peak hours, as well as on weekdays and weekends, to reflect a variety of driving situations under diverse traffic congestion scenarios.
To access the data, simply download the txt files available in the "Data" folder of this repository. The txt files contain time-speed profiles for each driving cycle, providing detailed temporal information about velocities, vehicle acceleration/deceleration parameters.
The code provided in this repository implements a data-driven framework for the generation and analysis of realistic driving cycles.
- Clone or download this repository to your local machine.
- Follow the instructions provided in the code comments to input the necessary parameters and paths to the data files.
- For MATLAB
- Set the data folder path
- Run "main.m"
- For Python:
- Set the data folder path
- In Google Colab
, Jupyter Notebook
, etc., upload provided "driving_cycle.ipynb" file and run the notebook cells.
- Note that in this study, segmentation is based on road characteristics. Segment sizes are given in the data folder as the "segment_sizes.mat" file. You can get different segment sizes by using different segmentation methods in your own work.
To run the code, you will need the following:
- Python with the necessary libraries installed. The required libraries for Python can be found in the "requirements.txt" file.
- MATLAB R2012a or later to run the MATLAB code.
-
Clone the repository:
git clone https://github.com/fesihkeskin/Driving_Cycle.git cd Driving_Cycle -
Install Python dependencies:
pip install -r requirements.txt
-
Ensure MATLAB R2012a or later is installed.
You can now run the code as described in the documentation.
You are free to use the code and data provided in this repository for research, analysis and development. If you use the code or data in this repository in your work, please cite our article:
@Article{Keskin2025,
author = {Fesih Keskin and Melih Yıldız and Bircan Arslannur},
journal = {Transportation Research Record},
title = {A Data-Driven Framework for Driving Cycle Generation and Analysis},
year = {2025},
number = {1},
pages = {2199-2212},
volume = {2679},
doi = {10.1177/03611981241260700},
url = {https://doi.org/10.1177/03611981241260700},
publisher = {SAGE Publications},
}Keskin, F., Yıldız, M., & Arslannur, B. (2025). A Data-Driven Framework for Driving Cycle Generation and Analysis. Transportation Research Record: Journal of the Transportation Research Board, 2679(1), 2199-2212. https://doi.org/10.1177/03611981241260700The code and data provided in this repository are licensed under the MIT License. See the
If you have any questions or feedback, please contact at: fesihkeskin@gmail.com