FP Growth - Max Number of Retries Parameter
Nevets
New Altair Community Member
Hi,
I am new to RapidMiner and have been doing the exercises in the book "Data Mining for the Masses" by Dr Matthew North.
For each exercise I have been trying to understand how each operator works and the underlying algorithms and mathematics.
In Chapter 5 "Association Rules" I couldn't quite work out why I was getting the results I was, whilst experimenting with FP Growth. The issue is around the
"max number of retries" parameter. The description for this parameter states:
This determines how many times the operator lowers min support to find the minimal number of item sets. Each time the minimal support is lowered by 20 percent. Range: integer; 2-+?; default: 15
After a lot of experimentation and offline calculations, I can only deduce the minimal support is lowered by 10% each time rather than the stated 20%. Also it seems as if the number of times it is lowered is 1 less than the value of this parameter. ie "max number of retries - 1" I'm guessing the first calculation at the "min support" value is being counted as a retry; or a coding bug where '<' is used instead of '<='.
Could someone verify the above, or if I am wrong, then how does this parameter work.
Thanks
I am new to RapidMiner and have been doing the exercises in the book "Data Mining for the Masses" by Dr Matthew North.
For each exercise I have been trying to understand how each operator works and the underlying algorithms and mathematics.
In Chapter 5 "Association Rules" I couldn't quite work out why I was getting the results I was, whilst experimenting with FP Growth. The issue is around the
"max number of retries" parameter. The description for this parameter states:
This determines how many times the operator lowers min support to find the minimal number of item sets. Each time the minimal support is lowered by 20 percent. Range: integer; 2-+?; default: 15
After a lot of experimentation and offline calculations, I can only deduce the minimal support is lowered by 10% each time rather than the stated 20%. Also it seems as if the number of times it is lowered is 1 less than the value of this parameter. ie "max number of retries - 1" I'm guessing the first calculation at the "min support" value is being counted as a retry; or a coding bug where '<' is used instead of '<='.
Could someone verify the above, or if I am wrong, then how does this parameter work.
Thanks
0
Answers
-
Hi,
well observed
I have created an internal ticket for the issues.
Regards,
Marco0
