"Extract eigenvector coefficients from PCA model"
Is is possible to extract the coefficients for computing the components of a PCA model?
I have a RM model that uses PCA to reduce 30 or so attributes down to 10 principal components. I need to compute these principal components inside a SQL database without calling out to RM, so I am trying to extract the coefficients that the model creates and save them into a database table.
The PCA-Eigenvectors view of the model looks a lot like an ExampleSet, but I have not been able to find something like a "ModelParameters2ExampleSet" node.
Thanks,
Keith
I have a RM model that uses PCA to reduce 30 or so attributes down to 10 principal components. I need to compute these principal components inside a SQL database without calling out to RM, so I am trying to extract the coefficients that the model creates and save them into a database table.
The PCA-Eigenvectors view of the model looks a lot like an ExampleSet, but I have not been able to find something like a "ModelParameters2ExampleSet" node.
Thanks,
Keith
0
Answers
this is currently not possible. However, we just added a new capability for (prediction) models indicating that a formula can be extracted from the model. Currently, only the SVM models are supporting this new feature and we plan to add other prediction models during the next months. Later on, we will probably extend this mechanism to the other models like PCA models.
Cheers,
Ingo