Just to give you the context, I’m working on predictive analytics (Linear regression). I would like to try Dimensionality Reduction using PCA but I don’t know how it works (the configuration on train and test data) so please provide an example workflow so that it would be helpful for me to understand what are the Nodes that I should use and how should I configure them.
Thank you very much,
you would want to use a combination of PCA Compute and PCA Apply nodes. The former will be trained on the training data and then the later will be used to apply the transformation to both the training and test data sets. Note, that you might want to normalise the data before calculation of PCA components.
There is a very nice example in this workflow that implements the analysis from the blog post Seven Techniques for Data Dimensionality Reduction.
This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.