BookmarkSubscribeRSS Feed
yunfeizhao100
Calcite | Level 5


We are using SAS Enterprise Miner 14.1. I have the below question for SAS Enterprise Miner. I am not sure where to get help, but I want to start from here. If you need more information, please let me know!

 

These questions are about to build a 0/1 binary target decision tree, where 1 is < 5% in data.

 

***Question 1, I open Interactive Tree window, I click Split Node, and only see 3-5 input attributes listed for use, and I am expecting nearly 100 input attributes? How can I get all my input attributes?

 

***Question 2, I run a Decision Tree node, and then I open Interactive Tree window. I see a few layers and nodes on the tree. I prune off part of the tree, and I want to Split Node from where I did the Prune Node. But I only see 3-5 input attributes, and I am expecting more input attributes. I prune off all nodes, and only keep the root node, then I save, close, and re-open the Interactive Tree, I might see more input attributes. What is happening here? What should I do?

 

***Question 3, My Decision Tree is typical 0/1, bad or not bad. But the Splitting Rule does not have Binary Target Criterion? What does it use for Binary Target? Interval Target Criterion, Nominal, or Ordinal?

 

***Question 4, I have millions of records, and nearly 100 input attributes. I basically use default values for Decsion Tree node, but change Use Input Once to Yes, change Maximum Branch to 4, leave maximum depth at 6, and click RUN, and it produces a tree. This tree gives me many leaves that only have a handful of records which is no use at all. So I change Leaf Size to ~2% of total records, now it produce a tree with only a single root node? What is happening here? What should I do?

 

***Question 5, What are typical property settings for Decision Tree node to build a 0/1 binary target tree (where 1 < 5% in the records), and to build a interval target tree, say outstanding balance? I have years of experience in using another commercial tool where I use default setting for everything, and results are as expected. I am fairly new to Eminer for the Decision Tree node and Data Source node, and I want to get started from expert advice here.

1 REPLY 1
yunfeizhao100
Calcite | Level 5

Thank you for helping me on above questions above!

sas-innovate-2024.png

Available on demand!

Missed SAS Innovate Las Vegas? Watch all the action for free! View the keynotes, general sessions and 22 breakouts on demand.

 

Register now!

How to choose a machine learning algorithm

Use this tutorial as a handy guide to weigh the pros and cons of these commonly used machine learning algorithms.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 1 reply
  • 551 views
  • 0 likes
  • 1 in conversation