Just a few questions. While fitting a decision tree, isn't a node split into two nodes only? Here, specifically for medium income group, with respect to age, the node has been split into four nodes, instead of two. Also, the two terminal nodes at the extreme left provide the same value of the dependent variable, which is "Bad" credit risk, following the majority class rule. But weren't the two nodes supposed to provide two different values of the dependent variable? Otherwise these terminal nodes would not have been created since they are not providing any different prediction from the node from which they got created (because the goodness of split value is low for the mother node here). Same goes for the terminal nodes at extreme right. Is all this due to the CHAID algorithm being used here?
Hello, sir; it was highly informative. Do you have any video about how to write interpretation with Decission Tree analysis? how do we ave to add image from SPSS etc.
Thanks for the video, it really helps me. Can i ask a question?. when i run the analyze, why there's only dependent variable node presented? while there's about 28 independent variables as an input?
Please check your sample size, might be low. secondly the chi square association might be weak and thirdly the number of samples in the parent node and child node please reduce sop that it can create a tree
My apologies for the delayed response, The idea of "influencer variable" is not clearly explained in SPSS documentation! Will get back to you once I find good information on this..
Hi Sir, thank you for this video. So if we use Decision tree we can identify the number of scorecard we need to developpe? after this wi use logistic regression of every segmentation?? thank you
you made a wonderful explanation of decision tree modeling via spss. thankyou
Thanks
Thank u sir .🙏😀
Just a few questions.
While fitting a decision tree, isn't a node split into two nodes only? Here, specifically for medium income group, with respect to age, the node has been split into four nodes, instead of two.
Also, the two terminal nodes at the extreme left provide the same value of the dependent variable, which is "Bad" credit risk, following the majority class rule. But weren't the two nodes supposed to provide two different values of the dependent variable? Otherwise these terminal nodes would not have been created since they are not providing any different prediction from the node from which they got created (because the goodness of split value is low for the mother node here). Same goes for the terminal nodes at extreme right.
Is all this due to the CHAID algorithm being used here?
CART model supports binary splits. however chaid supports multiple splits
This was a very helpful tutorial. Thank you so much!
Thanks Im glad you liked it.. All the best
Great breakdown, very good!
Glad you liked it!
Hello, sir; it was highly informative. Do you have any video about how to write interpretation with Decission Tree analysis? how do we ave to add image from SPSS etc.
in future i will make videos for this.
Very helpful sir 🔥
Thanks
great job, thanks man
Thanks
Thanks for the video, it really helps me. Can i ask a question?. when i run the analyze, why there's only dependent variable node presented? while there's about 28 independent variables as an input?
Please check your sample size, might be low. secondly the chi square association might be weak and thirdly the number of samples in the parent node and child node please reduce sop that it can create a tree
nice explanation ... Thank You 😊
Thanks 😊
hi would you mind sharing the link where you got the dataset from please?
docs.google.com/spreadsheets/d/1fsVX0ZL_-O5SCbBzieT6QSI_QBU-2CHp/edit?usp=sharing&ouid=110167476365142506887&rtpof=true&sd=true
In confusion matrix have you assumed probability as 0.5 as threshold for classifying as good and bad?
yes
What is the ''influence variable'' at the down? What is its implication in classification tree?
My apologies for the delayed response, The idea of "influencer variable" is not clearly explained in SPSS documentation! Will get back to you once I find good information on this..
Hi Sir, thank you for this video. So if we use Decision tree we can identify the number of scorecard we need to developpe? after this wi use logistic regression of every segmentation?? thank you
Running the decision tree is the first part then we need to create the score card. No need to run Logistic regression .
please sir
Not sure if you had a question?
Hi
Happy New Year!!