05-20-2017 03:25 AM
I wonder if someone can help me? I am using classification/machine learning algoithms using EM on my dataset then applying model comparison to see the accuracy of the algorithm. The purpose is to perform feature extraction and selection where possible.
The nodes I am using are HP SVM, Neural Network, HP Forest and Decision Tree
File Import >>>>Data Partition >>>> HP Forest, Neural Network, HP SVM, Decision Tree >>>> Control Point >>>> Model Comparison
Could anyone please tell me
1. How feature extraction works with these algorithms?
2. When I look at the results of these nodes , how can I find out which feature has been extracted,selected or rejected etc etc.
2 weeks ago
Not all of the nodes you indicated do variable selection. For example, the Neural Network uses all available inputs as does the HP SVM models, so it is important to consider doing variable selection prior to running the model. The Decision Tree does variable selection automatically -- the approach depends on the settings you choose -- while the HP Forest node does it optionally. In situations where you use the HP Forest node for variable selection, you would typically be recommended to run a second HP Forest node with only the important variables selected.
There is not a simple answer to your question but I would recommend you review the documentation available for all of these nodes in SAS Enterprise Miner by opening the application and clicking on Help --> Contents and then navigating in the panel on the left to the Node Reference. The help for the individual nodes is arranged by tab (Sample, Explore, Modify, Model, Assess). All of the modeling nodes you described are in the Model folder under the Node Reference.
Additional documentation is available at
You can read up on each of the modeling approaches and let us know if you have specific questions about something.
I hope this helps!