Building models with SAS Enterprise Miner, SAS Factory Miner, SAS Visual Data Mining and Machine Learning or just with programming

Adjusted Prior in Enterprise Miner 6.2.2 Please help? /something wrong Thank You

Reply
Occasional Contributor
Posts: 8

Adjusted Prior in Enterprise Miner 6.2.2 Please help? /something wrong Thank You

Hi ,
I have built a model in SAS entreprise Miner, but in the sample I used , the responds are over represented (1=0.5 and 0=0.5), so I have used the Decisions in EM and add the Adjusted Priors to correct it (1=0.0002 and 0=0.9998, because our responds are this much only), please see table below. when I connected this input sample data to decision tree it is not giving the variable importance in output window (but i can see in statexplorer) and giving the result in tree window like below. and when I connected the score node with score table , all customer responds are all 0s, not giving the proper output , please help me with this and please explain the step by step if possible if the responds are like this (0.0002) 
decision tree results      level 1    0%
                                      level 0  100%
                                      count 3042

level           count            prior       adjusted prior
1                 1521            0.5           0.0002
0                  1521            0.5          0.9998

Could you please help? Am I missing a step?
Many Thanks
Sathya

 

Super User
Posts: 10,516

Re: Adjusted Prior in Enterprise Miner 6.2.2 Please help? /something wrong Thank You

How many did you attempt to score?

From what you have shown I would not be surprised not to get any 1 with fewer than 10000 customers.

Occasional Contributor
Posts: 8

Re: Adjusted Prior in Enterprise Miner 6.2.2 Please help? /something wrong Thank You

actually my history table has 6000000 records with target 1s and 0s with 33 variables and responds were 1521(0.0002) that is why I am using sample node
and I attempted to score 600000 records/customers

please suggest me the best approach .
thanks you in advance
Ask a Question
Discussion stats
  • 2 replies
  • 300 views
  • 0 likes
  • 2 in conversation