topic Re: How to Match one Histogram/Distribution to another Histogram/Distribution?! in SAS Programming

How to Match one Histogram/Distribution to another Histogram/Distribution?!

hellohere — Sat, 09 Mar 2024 03:53:25 GMT

I get one histogram/distribution, which is good and close to Normal (see blow).

I get another one, which is not good and skewed heavily. (see below).

How to convert this histogram/distribution to the one above, but mapping/changing value?!

Code, or Logic?!

Re: How to Match one Histogram/Distribution to another Histogram/Distribution?!

ballardw — Sat, 09 Mar 2024 07:37:07 GMT

Are you requesting how to change the values of your variable(s) to a normal distribution or close to it? Or how to change the graph so the tails on the outliers to right do not appear?

If it is to change the appearance of the graph then you should provide the code you are currently using to create the graph so we can make informed suggestions. While a "picture is worth a thousand words" the picture can't actually show us things like the options used in code.

Re: How to Match one Histogram/Distribution to another Histogram/Distribution?!

hellohere — Sat, 09 Mar 2024 09:06:22 GMT

I need change the values of variable(s) to a perfect/theoretical normal distribution or close to it.
Better if Lamda Distribution due to non-zero.

By indexing, I know how to map A to B already.

Re: How to Match one Histogram/Distribution to another Histogram/Distribution?!

hellohere — Sat, 09 Mar 2024 09:07:58 GMT

Or simply how to generate the cumulative value of a perfect normal distribution, what function?

By indexing, I can map it.

Re: How to Match one Histogram/Distribution to another Histogram/Distribution?!

sbxkoenk — Sat, 09 Mar 2024 11:13:49 GMT

Hello,

You can standardize to a Given Mean and (given) Standard Deviation, but that does not make it 'normal' of course.
Base SAS Procedures Guide
STANDARD Procedure
Example 1: Standardizing to a Given Mean and Standard Deviation
https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/proc/p10clalffl60gsn1f2h17q0yf5k0.htm
Your best bet is probably the Box-Cox near-normality transformation.
Visit this page :
SAS/STAT 15.3 User's Guide
The TRANSREG Procedure
https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/statug/statug_transreg_details02.htm
... then scroll down to :
"The next example shows how to find a Box-Cox transformation without an independent variable. This seeks to normalize the univariate histogram."

Good luck,
Koen

Re: How to Match one Histogram/Distribution to another Histogram/Distribution?!

sbxkoenk — Sat, 09 Mar 2024 11:24:12 GMT

I think you will have to resort anyway to:

Box-Cox or Yeo-Johnson transformations.

See here:

The Box-Cox and Yeo-Johnson transformations for continuous variables
https://statisticaloddsandends.wordpress.com/2021/02/19/the-box-cox-and-yeo-johnson-transformations-for-continuous-variables/

I also have SAS-code for that Yeo-Johnson transformation ... but cannot spot it right now.

For Box-Cox code, see the post just above (with link to SAS doc).

Koen