Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Remove bar chart as permissible representation for aggregates #386

Closed
alexsb opened this issue Oct 17, 2017 · 10 comments
Closed

Remove bar chart as permissible representation for aggregates #386

alexsb opened this issue Oct 17, 2017 · 10 comments

Comments

@alexsb
Copy link
Member

alexsb commented Oct 17, 2017

Bar charts are not suitable to represent the median of a distribution due to within-the bar bias. Remove this option (add a dot plot instead, will create a separate ticket).

Here is an example for the matrix:

image

Here is an example for a group of a numerical column:

image

@sgratzl
Copy link

sgratzl commented Oct 18, 2017

depends how you see it. if it should a representation of an array I agree, if it should be an representation of a projection of an array to a single value, I disagree.

@mstreit
Copy link
Member

mstreit commented Oct 18, 2017

@sgratzl, I don't understand what you mean. What is your suggestion?

@sgratzl
Copy link

sgratzl commented Oct 18, 2017

I mean it is our decision how you wanna see and sell it. Both are valid options. I just wanted to bring up the other aspect, too.

@mstreit
Copy link
Member

mstreit commented Oct 18, 2017

OK. If we remove the single bar from the vis option, it also means that the column cannot be combined/stacked with other numerical columns.

@sgratzl
Copy link

sgratzl commented Oct 18, 2017

yes-ish. since it is based on the projection of 1D to 0D as mentioned before.

@alexsb
Copy link
Member Author

alexsb commented Oct 18, 2017

I don't agree. Even if you have a projection into a lower space, you represent an aggregate value in some way. That aggregate has uncertainty and variability associated with it, so the within-the-bar bias still applies.

See http://www.tandfonline.com/doi/pdf/10.1080/00031305.2016.1141706

I think for stacking aggregates, we should think of something fancier?

@mstreit
Copy link
Member

mstreit commented Oct 18, 2017

I get your point, but in Ordino users heavily use the stacking of columns where at least one is an aggregate.
Any ideas how 'something fancier' could look like?

@alexsb
Copy link
Member Author

alexsb commented Oct 18, 2017

Something like this:

image

Basics of a box plot, the darker areas show the possible positions of the mean for B and C.

Or this:

image

The black violin plots are the distribution of only that variable, the red ones show cumulative uncertainty.

@sgratzl
Copy link

sgratzl commented Oct 18, 2017

-> should be moved to #201

@mstreit
Copy link
Member

mstreit commented Oct 24, 2017

solved with #362 and #401.

image

@mstreit mstreit closed this as completed Oct 24, 2017
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

3 participants