It is not easy to work with the metadata files. I have been thinking about a small program that would go like this:
Give path to datafile -> buildmetadata()
If metadata file already exists, ask if you want to remake. Y/N
Outputs preliminary-meta-datafile.csv
Reads preliminary-meta-datafile.csv
loop over all variates {
1. print the name of the variate
2. Do you want to keep this variate? Y/N (No, go to next variate, Yes continue)
3. This looks like a _ <- best guess at broad (binary, categorical, numerical) variate type. Is that correct? Y/N
4. If relevant, plot the variate KDE/histogram
5. Ask detailed questions like max/min value, number of categories etc (specific to each variate) and updates relevant lines of text.
}
Save the updated metadata file as meta-datafile.csv
It would not be hard to make, although it would take some time and testing, but it is also not an urgent feature.
It is not easy to work with the metadata files. I have been thinking about a small program that would go like this:
It would not be hard to make, although it would take some time and testing, but it is also not an urgent feature.