Refactor columns and features in Azure Machine Learning

Question

Is there any way I can make my dataset features in Azure ML into something else than what it already is?

I found a dataset of the Titanic ship in the sample datasets which I would like to work with but all of my columns are either a numeric feature or string feature, but I would like to categorize these. Also is there any possibility to rename the columns within my model so it’s more descriptive than what I initially got? I have no clue what SibSp means for instance.

Answer 1

What you are doing is essentially recreating this experiment made by Raja Iqbal for the Titanic dataset. I recommend you check that out here: http://gallery.cortanaintelligence.com/Experiment/Tutorial-Building-a-classification-model-in-Azure-ML-8?share=1

To answer your question, the module you can drag to your canvas in order to make the features into categories; is the Edit Metadata module where you select the columns you want and change the “unchanged” into “Make categorical” within the Categorical-properties pane like in the image below:

You can also use the same module to make better sense from your columns by giving them a different column name. SibSp means SiblingSpouse like I have renamed it to in the image below:

And at last you can assign the targeted value (survived) and make the field into a label for ease of use.