Date in auto model

data1maths

Hello everyone,

I noticed that every time i run an auto model with a dataset having date as a variable(column) it doesn't include it in the prediction or the classification, even if check its format as date-time format, but nothing works.

 

I hope that you can help me with this problem, because i really need the model to take the date colunm in consideration.

Thank you.

IngoRM

Hi,

 

You are right, date columns are currently not used in Auto Model.  But here is what you can do (and typically get better results by doing this):

 

1. Go to the Process view
2. Build a simple data prep process there where you load your data set (containing the date column), extract some meaningful information (see next point), and store the final result in your repository again.  Run this process.
3. Use the operator "Date to Numerical" and extract meaningful new features from your date column.  Those could be things like "Month of the Year", "Day of the Week", "Quarter", "Calendar Week" etc.  Only you know what makes most sense for your application.
4. Run Auto Model on this transformed data set with the extracted information.

Hope this helps,

Ingo

IngoRM

Hi,

 

You are right, date columns are currently not used in Auto Model.  But here is what you can do (and typically get better results by doing this):

 

1. Go to the Process view
2. Build a simple data prep process there where you load your data set (containing the date column), extract some meaningful information (see next point), and store the final result in your repository again.  Run this process.
3. Use the operator "Date to Numerical" and extract meaningful new features from your date column.  Those could be things like "Month of the Year", "Day of the Week", "Quarter", "Calendar Week" etc.  Only you know what makes most sense for your application.
4. Run Auto Model on this transformed data set with the extracted information.

Hope this helps,

Ingo

data1maths

Thank you so much Mr Ingo for your help.