How to a use Auto Model for data that I have already split into train and test?

kzhao25

I am trying to solve an imbalanced binary classification problem using a model to predict the minority class (stroke victims).  I used oversampling on the training data to make synthetic instances of stroke cases so that I could address the data imbalance issue.

However, I kept the test data as its normal imbalanced distribution rather than oversampling that too because I want to test my model on the real-world distribution.  I would like to use RapidMiner's automodel feature, but every time I try to use it then it just splits my training data into another train-test split and does its own thing.

How do I use Auto Model while specifying the data that those models should be trained on and the data that it should be tested on?

Marco_Barradas

Hi @kzhao25,

You'll need to save Automodel output or you best model and then you can reuse the apply model operator to score your hold out data.

Scoring demo | RapidMiner Studio

Machine learning - classification | RapidMiner Auto Model