search

opendp / dp-wizard

Web application that makes data releases that satisfy differential privacy using the OpenDP Library
MIT License
0 stars 1 forks source link

Make notebook with plots for columns #152

Closed mccalluc closed 1 week ago

mccalluc commented 2 weeks ago

Dropped the expected-script test: when it was short, it was useful, but now it just means that every change requires a corresponding update to the test: Better to have finer grained unit tests that catch problems earlier.

In the notebook template, added lots of comment blocks that will be turned into markdown cells. The # + and # - are the jupytext light format.

This is a change, but I think that having the executed notebook be the real release from the perspective of this application makes sense: That's the best way of making sure the code in the notebook does what we think it should do.

For reviewer:

ekraffmiller commented 1 week ago

Looks good, I started from scratch with the new repository and downloaded a notebook with histograms created. One thing I noticed, if I change the number of bins using the arrows, then the bins appear out of order in the histogram plot. This happens in the app and in the notebook. For example,

Screenshot 2024-11-15 at 11 32 05 AM

mccalluc commented 1 week ago

bins appear out of order in the histogram plot

Good catch: These strings are being sorted alphabetically, so "10" comes before "2". I think the best way to fix this is to pull out the first number of a pair and sort, probably on the python side, rather than trying to parse the string inside polars. Will file a new issue.

mccalluc commented 1 week ago

should I test the generated python script, or is that still a work in progress?

Not really necessary right now. There's going to be more work on these.

mccalluc commented 1 week ago

Going ahead and fixing the sorting...

mccalluc commented 1 week ago

Now also