It’s World Statistics Day! To honor the theme of the day, the JMP User Community is having conversations about the importance of trust in statistics and data. And we want to hear from you! Tell us the steps you take to ensure that your data is trustworthy.
Choose Language Hide Translation Bar
Highlighted

Johnson transform

I am trying to find a normalizing transform for my data so that I can model my data with the normal distribution.

I am using Analyze-Distribution by [my product], for all six of my Y variables.

I have two informal criteria:

1. AICc within 5 of the lowest AICc for any distribution for that Y var and product.

2. Shapiro-Wilk W test Prob<W >0.05.

This is telling me that the Johnson SU is overall definitely the best choice.

My question is, how do I tell JMP Fit-Model to use the Johnson SU distribution?  I don't see the distribution option that should be there!  ARRGGGHH.

I am not JSL literate yet.  If I have to do the transforms manually (in Excel), is it as simple as calculating ASINH of each value then copying that into a data table?  Then, on any conclusion, I have to transform back using SINH?  Or is it more complicated?  Do I have to include estimates of each of the four JSU parameters into the ASINH and SINH to do it properly?

1 ACCEPTED SOLUTION

Accepted Solutions
Highlighted
louv
Staff (Retired)

Re: Johnson transform

When you are in the distribution platform you can click on the red-triangle and save the transformed data back to the table and model that transformed normal response.

8058_Screen Shot 2015-02-18 at 7.22.36 PM.png

View solution in original post

1 REPLY 1
Highlighted
louv
Staff (Retired)

Re: Johnson transform

When you are in the distribution platform you can click on the red-triangle and save the transformed data back to the table and model that transformed normal response.

8058_Screen Shot 2015-02-18 at 7.22.36 PM.png

View solution in original post

Article Labels

    There are no labels assigned to this post.