Solved: Re: X-axis format

Ted · Jun 28, 2019 12:53 PM

How to divide the X-axis into equal parts (SEE picture)?
P.S. X-axis is not interactive in the "Y by X platform"

julian · Jun 28, 2019 01:37 PM

Mosaic plots (which is the default visual you will get in Fit Y by X when you use a categorical X and categorical Y) do not divide the x-axis evenly unless there is an equal number of observations in each of the categories represented by the x-axis. You can see the documentation for mosaic plots here. There isn't a way to disable this scaling in Fit Y by X because it's a feature of the plot itself. And, it's a good thing! Because of that scaling, you can see the marginal distribution of your X-variable, which is quite useful to see in addition to the conditional distribution of Y for each X category.

If you absolutely need a plot that sizes the X-axis uniformly irrespective of the distribution of X, it won't be a mosaic plot. But, you can make such a plot in Graph > Graph Builder, by placing your Y variable in the Y Role, and your X variable not in the X-Role, but in the Group X role at the top. Then, toggle the mosaic plot at the top to make a mosaic-like plot.

I hope this helps,

@julian

View solution in original post

Mark_Bailey · Jun 28, 2019 01:37 PM

Sorry - I was mistaken that you could turn off the proportional spacing in the Contingency platform. (I never wanted to turn it off.) I must be thinking of Oneway, another platform under Fit Y by X.

julian · Jun 28, 2019 01:37 PM

Hi @Ted,

Mosaic plots (which is the default visual you will get in Fit Y by X when you use a categorical X and categorical Y) do not divide the x-axis evenly unless there is an equal number of observations in each of the categories represented by the x-axis. You can see the documentation for mosaic plots here. There isn't a way to disable this scaling in Fit Y by X because it's a feature of the plot itself. And, it's a good thing! Because of that scaling, you can see the marginal distribution of your X-variable, which is quite useful to see in addition to the conditional distribution of Y for each X category.

If you absolutely need a plot that sizes the X-axis uniformly irrespective of the distribution of X, it won't be a mosaic plot. But, you can make such a plot in Graph > Graph Builder, by placing your Y variable in the Y Role, and your X variable not in the X-Role, but in the Group X role at the top. Then, toggle the mosaic plot at the top to make a mosaic-like plot.

I hope this helps,

@julian

Evan_Morris · May 19, 2022 05:10 AM

Not sure if this is still an issue in the newer versions of JMP, but the problem I have is the following:

For the X by X contingency I can't adjust the y-axis which looks lousy for low frequency issues, so I go to the graph builder to use the mosaic. But there the X-axis can't be adjusted. Doing the Group X binning works ok but then I lose the built in chi-sq (I can solve it in x by x and then bring it back over but that's a little clunky). I don't really care if I do it graph builder or x by x but I need it somewhere where I can get a clean, easy to read graph to go with my chi-sq.

Maybe just make the y-axis scalable in the normal contingency plot?