Rapid miner cannot show decision tree for sample size above 100 .

fatimidveil
fatimidveil New Altair Community Member
edited November 5 in Community Q&A
hi ,everyone ,my data set consist of 1150 entities with 49 variables . i am using  rapid miner 9.4 free version .
My rapid miner show decision tree and working smoothly with sample size 100 but when, i increase my sample size from 100 than my software didn't show complete  tree ,only show one variable with class .
i hope you people will help me out from this difficult situation .
Tagged:

Best Answer

  • varunm1
    varunm1 New Altair Community Member
    edited October 2019 Answer ✓
    So, based on your process I can see that you are using an ID3 tree builder. I can see that it is building a decision tree based on the PTM attribute and there is nothing wrong with this. One thing I can say is that the PTM attribute is 100 percent correlated with the response variable (college), so there is no need for your tree to make decisions based on other attributes, so it might not use them for tree building. Generally, if some attribute is 100 percent correlated to the response variable, we will be a bit cautious and see if this is not some replica of response that we are trying to predict. 


    Please inform if you need more information.

Answers

  • varunm1
    varunm1 New Altair Community Member
    Hello @fatimidveil

    Looks like your tree is getting pruned. Can you inform the perfomance values of your tree, if you are trying to get one? One more thing is, can you uncheck pruning in decision tree parameters then run the process and see if your are getting any tree now? 
  • fatimidveil
    fatimidveil New Altair Community Member
    no after uncheck the pruning i am getting the name tree :-(

  • varunm1
    varunm1 New Altair Community Member
    Thanks for your response. Is it possible for you to provide the process you built by exporting it into .rmp file (FILE --> Export Process) and attach it here in thread. If you were able to provide data that would be helpful to reproduce your result and give you some insight.
  • fatimidveil
    fatimidveil New Altair Community Member
    this is my process
  • fatimidveil
    fatimidveil New Altair Community Member
    this is my data file it contain 200 sample size 
  • varunm1
    varunm1 New Altair Community Member
    edited October 2019 Answer ✓
    So, based on your process I can see that you are using an ID3 tree builder. I can see that it is building a decision tree based on the PTM attribute and there is nothing wrong with this. One thing I can say is that the PTM attribute is 100 percent correlated with the response variable (college), so there is no need for your tree to make decisions based on other attributes, so it might not use them for tree building. Generally, if some attribute is 100 percent correlated to the response variable, we will be a bit cautious and see if this is not some replica of response that we are trying to predict. 


    Please inform if you need more information.
  • fatimidveil
    fatimidveil New Altair Community Member
    thankyou so much for your valuable response ...
  • fatimidveil
    fatimidveil New Altair Community Member
    looking forward to your suggestion in future thank you so much 

Categories