Feature Request: Can you add a comparison mode for numerical data?

[EED85]

Hi everyone,

This looks great and very promising!

The comparison of categorical data and its shifts is presented very well. For numerical data, it would be helpful to display changes using defined measures such as average, maximum, minimum, quantiles, etc.

Is something like this planned?

Thanks,
Eric

[MariusMerkleQC]

Hi @EED85, thanks for your suggestion! We have already considered this, but there weren't enough use cases yet. Could you tell us more about your use case?

The comparison of categorical data and its shifts is presented very well.

I assume you're talking about the summary here. Why would you say that comparisons of categorical data are presented better than comparisons of numerical data?

[MariusMerkleQC]

One more question came up in a discussion with @EgeKaraismailogluQC: when you talk about the measure, for example the maximum, do you mean

1. the maximum of one data frame vs the maximum of the other data frame (-> in this case, no (valid) primary key is required), or
2. the maximum deviation within rows matched via the primary key

Let's say we have two dataframes

Left

PK	a
"A"	1
"B"	6

Right

PK	a
"A"	3
"B"	7

Matched

PK	a_left	a_right	a_diff
"A"	1	3	2
"B"	6	7	0

In 1), the maximum would change from df_left["a"].max() == 6 -> df_right["a"].max() == 7
In 2), the maximum would be df_matched["a_diff"].max() == 2

Which of the two are you interested in?

[martinrohbeck]

Thanks for opening this issue, @EED85 and @MariusMerkleQC . I would like to 2nd this request.

What I'd be interested in is your first case, i.e. see how the min (1 -> 3), max (6->7) and mean/median (3.5->5) values change between the column a.

[MariusMerkleQC]

Makes sense. Are you interested in the same, @EED85?