This tool was developed by Filip Arnaut from the Institute of Physics Belgrade, University of Belgrade, as part of the research paper titled:
"Multi-Class Machine Learning Anomaly Detection of Ionospheric VLF Data: Dataset and Web-based Framework "
The tool enables users who have downloaded .mat (MATLAB) ionospheric VLF amplitude data from Waldo World to quickly and easily convert the data into a CSV file compatible with the TRAINSET format- a browser-based tool for data labeling/classification.
How it works
1. Upload Your File
Click the “Choose File” button and upload your .mat (MATLAB) file containing VLF amplitude data. The file should contain the standard metadata and data array typically provided by Waldo World.
2. Set Downsampling (Optional)
Choose a downsampling factor if you'd like to reduce the number of data points. For example, selecting 5 keeps every 5th data point. Set it to 1 if you want to keep all data.
3. Convert and Download
Click “Convert and Download CSV.” The tool will extract the data, generate timestamps based on the metadata in the file, and export the dataset in TRAINSET format. This includes:
- series: pre-filled with "series_a"
- timestamp: in ISO8601 format
- value: the VLF amplitude value
- label: set to "Normal" by default
4. Use the Output
The generated CSV can be loaded directly into the TRAINSET tool for further processing.
Its use is free and encouraged for anyone interested.
For further information or inquiries, please contact: filip.arnaut@ipb.ac.rs
Sample input and output files are available at: https://github.com/arnautF/mat-to-csv-web