Showing values for bar plots in TableReport
Problem Description
Hello, It would be great if we can see the value of a horizontal bar , or a vertical bar in the TableReport plot. Currently the situation is the following:
- For vertical bars, we cannot know the exact value of a bar (X and Y), especially when we have a single tick label in the axes, with large values .
- For horizontal bars, we can see the percentage of each bar, but not the value as well.
Suggestion:
- Showing more info about each bar
on_hover, similar to what is used in other libraries likeplotly. - This way the layout of TableReport will remain the same, but the info will be shown
on_hoveronly
Below is an example of the current TableReport plots.
Feature Description
- Adding an
on_hoverevent to the plots - This may open the doors to simplify the implementation of "tailored" plots in TableReport
Alternative Solutions
Using off-the-shelf solutions may solve it ex.plotly, but this will add a new dependency for skrub , and it is not straight forward to execute javascript inside templates (currently in skrub, the plots exist inside template tags)
Additional Context
No response
Hey @alamehm, thanks a lot for the suggestion!
I agree that the current version of the plots in the TableReport could be improved (having only one tick in some cases should be fixed, for example).
I think that having the exact value of a bar could be quite useful, and straightforward to implement. It could also be a kwarg or a configuration parameter for convenience.
I am not keen on adding a dependency on plotly for that, and in general I'm afraid that replacing the current plots with a plotly version would likely introduce a lot of issues.
Tailored plots are a cool idea that I remember from the workshop, and while there is some refactoring to be done, it shouldn't be too difficult.
So, I like the suggestion but I am not convinced that implementing it with plotly would be the best way of doing it 🤔
@rcap107, I agree with you about the issues of adding a new dependency.
I try to open an Pull Request for plotting the value on the histogram.