Transformations for converting data into normaly distributed

2195
1
Jump to solution
02-27-2012 03:03 AM
YatharthMathur
New Contributor
During analysis of data from a histogram how can i decide what transformations to use ...
Also for some entries histogram does not has an option of transformations .[ATTACH=CONFIG]12245[/ATTACH]
0 Kudos
1 Solution

Accepted Solutions
EricKrause
Esri Regular Contributor
You have a data value equal to 0, so the logarithmic and Box-Cox transformations won't work (the data has to be all positive).  And you have data bigger than 1, so ArcSin won't work (all data has to be between 0 and 1). 

There's also a Normal Score transformation (NST), but you only have 11 points, and that isn't near enough to do a valid NST. 

With only 11 points, don't use kriging.  We say that you should have at least 30 points to even attempt kriging, but you really should have 100 or more.

View solution in original post

0 Kudos
1 Reply
EricKrause
Esri Regular Contributor
You have a data value equal to 0, so the logarithmic and Box-Cox transformations won't work (the data has to be all positive).  And you have data bigger than 1, so ArcSin won't work (all data has to be between 0 and 1). 

There's also a Normal Score transformation (NST), but you only have 11 points, and that isn't near enough to do a valid NST. 

With only 11 points, don't use kriging.  We say that you should have at least 30 points to even attempt kriging, but you really should have 100 or more.
0 Kudos