ENH - Add a new tab to the TableReport to show the global dataframe stats

[rcap107]

Problem Description

The TableReport in its current form includes statistics at the column level, such as dtype, number of nulls, number of unique values, numeric statistics etc.

There are multiple dataframe-level statistics that would be interesting and that could be added to a separate tab, such as:

• Shape of the dataframe (currently shown at the bottom of the preview screen)
• Estimated size in memory (will be added in Skrub Add Estimated Memory Usage to TableReport #2153)
• Presence of null values
• Number of columns by dtype (numeric columns, string columns, datetime columns etc)

Additionally, it may be interesting to have a "Description" field that can be filled by users so that it is easier to share information about the report (this would be in addition to the title).

Feature Description

This could be implemented as a separate tab, possibly to the left of the Preview, while keeping the preview tab as the default.

Alternatively, this could be added in the "Stats" tab, but I am not sure where it would fit given that the stats are already taking up the entire width of the screen.

Alternative Solutions

No response

Additional Context

No response

[sabasiddique1]

Hi @rcap107, I’d be happy to work on this if contributor help is welcome. I was thinking of starting with a new Overview tab for dataframe-level stats, while avoiding overlap with the memory-usage work in #2153.

[jeromedockes]

thank you @sabasiddique1 for proposing! before starting working on it first we need to decide if we want to add this (most of the information is already somewhere in the report, a new tab adds clutter and may cause the scrolling of the tabs to kick in) and what it would look like

[rcap107]

Discussed this IRL

Overall, there isn't a lot of enthusiasm about adding this as a separate tab. It was instead suggested to have a separate table that appears in the preview tab, similarly to how we are handling the display of distributions for each column.

As a more general note, we might be trying to add too many features to the report, which is something that we may want to avoid.

[jeromedockes]

As a more general note, we might be trying to add too many features to the report, which is something that we may want to avoid.

based on the IRL discussion I will close this for now. we can always reopen