Dark Mode
Geophone Sensor HAR Feature Dataset
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
Vibration data collected using a geophone sensor designed specifically for the analysis of human movement patterns. This data product serves to identify and differentiate between three fundamental human activities: walking, running, and standing still (waiting). The raw vibration signals have been processed to derive 12 statistical and frequency-domain features, segmented into clean 3-second time windows for ready use in machine learning applications. The segmentation ensures high-quality features suitable for developing robust activity recognition models.
Columns
The dataset includes 14 columns detailing time windows, extracted features, and activity labels:
- timestamp: Provides time context for the data readings.
- mean: The average value of the signal over the observation window.
- top_3_mean: The average value derived from the top three signal readings in the window.
- min: The smallest value recorded in the signal during the time window.
- max: The largest value recorded in the signal during the time window.
- std_dev: A statistical measure of signal variability, indicating deviation from the mean.
- median: The middle value of the signal, dividing the data into two equal halves.
- q1: The first quartile, representing the 25th percentile of the signal data.
- q3: The third quartile, representing the 75th percentile of the signal data.
- skewness: Measures the asymmetry of the signal distribution within the window.
- dominant_freq: The frequency exhibiting the highest power, useful for understanding signal periodicity.
- energy: The total energy of the signal, calculated as the sum of squared signal values.
- activity: The categorical label indicating the specific human activity (e.g., walking, running).
- name: An identifier associated with the individual performing the activity.
Distribution
The data is delivered as a CSV file (
geophone-sensor-data.csv) and contains a total of 1,800 rows across 14 columns. The entire file size is approximately 255.05 kB. The data structure segments activities into 3-second time windows, with each individual contributing 120 rows per activity type. The data shows excellent integrity, with no missing or mismatched values reported across the fields. Updates are expected on a weekly basis.Usage
This data product is ideal for training high-performance machine learning models focused on activity recognition. Specific use cases include user identification based on gait or movement signature, developing advanced classification algorithms for sensor data, and general time-series analysis applications where feature detail is paramount.
Coverage
The data reflects measurements taken over a short duration, roughly spanning 26 December 2024 to 27 December 2024. The measurements were sourced from five distinct individuals: Furkan, Enes, Yusuf, Alihan, and Emir. The scope of activities is strictly limited to walking, running, and waiting (standing still).
License
CC0: Public Domain
Who Can Use It
- Data Scientists: For feature engineering and classification model development.
- Machine Learning Engineers: To create and benchmark models for real-time human activity monitoring systems.
- Academics and Researchers: Studying biomechanics, sensor reliability, or developing novel signal processing techniques.
- Engineers: Working on embedded systems that require low-latency physical activity detection.
Dataset Name Suggestions
- Geophone Human Activity Feature Set
- Vibration-Based Human Movement Data
- Sensor Data for Step Identity Classification
- Geophone Sensor HAR Feature Dataset
Attributes
Original Data Source: Geophone Sensor HAR Feature Dataset
QUALITYLoading...