CODEWITHSUNDEEP
azuremachine-learningazure-machine-learning-service
lja
lja

Reputation: 1

How to use historical data set for training and prospective data set as input for prediction in Azure Machine Learning

Background information: I did an Data Mining experiment where I used historical data of customer purchases as case table for my mining structure. The second data set (prospective buyers) is used for testing.

Now I want to implement the same scenario in Azure Machine Learning (Studio). However, I could not figure it out how I can use one data set to be used for training and a different data set to be used for testing.

Furthermore, I'd like to ask if it is possible to use a data set for training the model but after deploying the model to a web service, to limit the input fields to certain columns?

The historical data set contains 12 columns that I want to use for training the model. However, I want only 9 of those columns to be required as input when testing via the deployed model.

I hope I made myself clear and that everything is understandable. If not, please ask me anything you want.

Kind regards, lja

Upvotes: 0

Views: 857

Answers (1)

BenjaVR
BenjaVR

Reputation: 550

However, I could not figure it out how I can use one data set to be used for training and a different data set to be used for testing.

You can do it like this: enter image description here

Please note both data sets should have the same columns!

The historical data set contains 12 columns that I want to use for training the model. However, I want only 9 of those columns to be required as input when testing via the deployed model.

The model (and the generated web service) needs the columns you trained with to feed the model. If those other 3 columns are not required, just leave it empty.

If you have your own application consuming the web service, you can just ask the input fields you want, and send empty values behind the scene.

Upvotes: 1

Related Questions