Hello,
I have a couple of questions around using Model Studio for creating unsupervised models. I am currently using the "cluster object" in VA to make a segmentation, using k-means. I would like to move it to Model Studio. For the models such as a logistic regression, I have an option on my VA screen to click on "create pipeline" button to add my model to a current project in Model Studio. However, I think I cannot do that for cluster analysis in VA, would that be correct?
So, I wanted to start a new project from scratch in Model Studio to replicate what I have done in VA, but there it is required to select a target variable, where I do not have one because it is a segmentation project, not a predictive model. So, I have chosen an ID variable as the target, just to be able to create a pipeline. So, my second question is, what would be the suggested target variable selection process for a k-means segmentation project, as I suspect my method is correct?
Finally, I can add "clustering" node to the pipeline after the target variable selection but I am interested in adding several clustering nodes to the pipeline and making a comparison of its results. I Therefor, I'd like to use Model Comparison node but clustering node and model comparison nodes cannot be used together without adding a supervised learning node, I believe. So. my third question is, what should I do to create few different cluster nodes and compare their results?
Hope that makes sense! Thanks for the support.
