The ENERTALK Dataset, 15 Hz Electricity Consumption Data from 22 Houses in Korea [paper] [dataset]
This repository contains jupyter notebooks for plots in the ENERTALK dataset paper. In order to visualize figures using notebooks, the 'ENERTALK' dataset must be downloaded first.
Listed in requirements.txt. Install with pip install -r requirements.txt
After acquiring the data and running the setup scripts your folder structure should look like
PROJECT_PATH/
enertalk-dataset/
00/
20161101/
20161102/
...
01/
...
nilmtk-converter
notebooks
tools
This folder contains nilmtk converter and metadata generator for nilmtk conversion.
metadata/
- Metadata generated by metadata_generator.py
metadata_generator.py
- Metadata generator in accordance with nilm_metadata
convert_enertalk.py
- data converter for nilmtk
The visualization codes are used for the following figures.
Fig04.ipynb
- Figure 4: Data snippets of two houses.
Fig04_reduced.ipynb
- Downsampled version of Figure 4.
Fig05.ipynb
- Figure 5: Data snippets of two houses - measurements plots of six appliances.
Fig06.ipynb
- Figure 6: Snippets of six refrigerators from six different houses.
Fig07.ipynb
- Figure 7: Plots of power consumption measurements for sampling rate of 1/6 Hz, 1 Hz, 15 Hz.
Fig08.ipynb
- Figure 8: Hourly distribution of average power consumption (house 00)
Fig09.ipynb
- Figure 9: Top four cluster centers found with adaptive K-means clustering on the normalized data.
Fig10.ipynb
- Figure 10: Box plot of daily on-ratio over the 22 houses.
Fig11.ipynb
- Figure 11: Heatmap of daily observation ratio for the 22 houses.
Fig12.ipynb
- Figure 12: Average occurrence count of gaps for the 22 houses.
Fig13.ipynb
- Figure 13: Observation ratio of House 00.
Also, 'notebooks' folder contains the simple test code for the nilmtk
nilmtk_test.ipynb
- Simple loading h5 file generated by nilmtk converter