ShieldX/Rash-Driving-Detection-on-Bikes-for-ML

Dataset Card for Rash Driving Detection on Bikes Using Mobile and Sensor Data

This dataset is designed to aid the detection of rash driving behavior on bikes using data collected from mobile and sensor-based systems. It includes sensor readings such as accelerometer values, orientation (azimuth, pitch, roll), and speed, with labels indicating whether the riding behavior is classified as rash or not.

Dataset Details

Dataset Description

This dataset consists of sensor data from bikes equipped with mobile sensors, specifically aimed at detecting rash driving behavior. The dataset contains the following features:

• Accelerometer Data: Accelerations along the x, y, and z axes (ax, ay, az).
• Sensor Orientation: Azimuth, pitch, and roll values describing the orientation of the bike.
• Label: A binary target variable (is_rash) indicating whether the riding behavior is considered rash (1) or not (0).

This dataset is intended for training machine learning models and anomaly detection systems focused on road safety and rider behavior analysis.

• Curated by: Rohan Shaw
• Shared by: Rohan Shaw
• License: MIT

Dataset Sources

• Repository: https://huggingface.co/datasets/rash-driving-detection-on-bikes-for-ml

Uses

Direct Use

This dataset is primarily intended for:

• Rash driving detection: Using sensor data to classify whether riding behavior is rash or safe.
• Machine learning model training: The dataset can be used to train models for real-time behavior classification.
• Anomaly detection: Detecting outliers or unsafe behavior patterns that could indicate dangerous driving conditions.

Out-of-Scope Use

• Non-vehicular data analysis: This dataset is specifically for bikes, and might not generalize well to other types of vehicles.
• Privacy-sensitive applications: Although the dataset includes location data, it is anonymized and should not be used in applications that aim to track individuals or breach privacy.

Dataset Structure

The dataset contains the following fields:

Field	Type	Description
timestamp	Integer	Timestamp of the data entry
ax	Float	Accelerometer value along the x-axis
ay	Float	Accelerometer value along the y-axis
az	Float	Accelerometer value along the z-axis
azimuth	Float	Azimuth angle of the bike
pitch	Float	Pitch angle of the bike
roll	Float	Roll angle of the bike
is_rash	Integer	Rash driving label (1 = rash, 0 = safe)

Dataset Creation

Curation Rationale

The dataset was created to provide a reliable and robust dataset for detecting rash driving behavior in bikers based on sensor readings. It was motivated by the need to improve road safety and to enable real-time monitoring of rider behavior.

Source Data

The data is collected from mobile sensors attached to bikes, capturing readings such as accelerometer data, orientation, and geospatial information. These readings are recorded at regular intervals while the bike is in motion.

Data Collection and Processing

The data was collected using mobile sensors that measure accelerations, orientation (pitch, roll, azimuth), and speed. The raw sensor data was processed and cleaned to remove outliers and erroneous readings. The data is anonymized to protect personal information, and location data is generalized for privacy.

Who are the source data producers?

The data was collected by Rohan Shaw as part of an ongoing study on road safety and driver behavior.

Annotations

Annotation process

The data is annotated based on observed riding behavior. Riding behaviors that involve erratic or unsafe patterns, such as sudden accelerations or sharp turns, are labeled as "rash" (1), while stable and controlled riding is labeled as "safe" (0).

Who are the annotators?

The annotations were performed by Rohan Shaw with a focus on safety and road behavior.

Personal and Sensitive Information

The dataset does not contain any personally identifiable information or sensitive data. While location data is provided (latitude, longitude), it is generalized to prevent the identification of specific individuals.

Bias, Risks, and Limitations

Bias

As the dataset is primarily focused on bikes, the model trained on this dataset may not perform well for other types of vehicles. Additionally, the dataset may be biased towards certain road conditions or geographic areas where the data was collected.

Risks

The dataset could potentially be used in ways that may affect privacy (e.g., overemphasis on location data). Efforts have been made to anonymize sensitive information, but users should exercise caution when using the dataset in sensitive applications.

Limitations

The dataset might not capture every type of rash driving behavior or all possible road conditions. It's important to consider that environmental factors (e.g., weather, road type) may also influence riding behavior, which is not included in this dataset.

Recommendations

Users are encouraged to:

• Evaluate models in diverse environments: Ensure that the model trained on this dataset is tested across different road conditions, locations, and rider demographics.
• Consider privacy: While this dataset is anonymized, avoid using it for applications that require detailed personal information or geo-tracking.

Citation

If you plan to cite this dataset, here’s an example:

BibTeX:

@misc{rash-driving-dataset,
  author = {Rohan Shaw},
  title = {Rash Driving Detection on Bikes Using Mobile and Sensor Data},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/rash-driving-detection-on-bikes-for-ml}},
}

APA:

Rohan Shaw. (2025). Rash Driving Detection on Bikes Using Mobile and Sensor Data. Hugging Face. https://huggingface.co/datasets/rash-driving-detection-on-bikes-for-ml

Glossary

• Accelerometer: A device that measures proper acceleration, the rate of change of velocity.
• Azimuth: The angle between the reference direction (usually North) and the line connecting the sensor to the object.
• Pitch: The angle of the sensor relative to the horizontal plane (up/down tilt).
• Roll: The angle of the sensor relative to the vertical axis (side-to-side tilt).

More Information

For more details or usage examples, please refer to the official repository or contact Rohan Shaw.

Dataset Card Authors

• Rohan Shaw

Dataset Card Contact

For inquiries, please contact: rohanshaw.dev@gmail.com