Retrieve the associated scikit clustering model with DSS

jianhao18
Level 1
Retrieve the associated scikit clustering model with DSS

Hi,

Following the guide here API for interacting with saved models โ€” Dataiku DSS 7.0 documentation,  I have been able to retrieve the scikit models with DSS (for prediction models - regression/classification). However, this does not work on clustering models as shown in the screenshot, the get_predictor() method fails with the following error

 

 

 

<class 'TypeError'>: preprocessing_steps() got an unexpected keyword argument 'with_target'

 

 

 

Has anyone faced this problem before? Would really appreciate your help on this, I have attached the screenshot below. Thanks!

FYI: I am using DSS version 7.0.3

tmp.jpg

tmp2.jpg

0 Kudos
3 Replies
tgb417

@jianhao18 ,

Welcome to the Dataiku Community.

I have not faced this specific problem in the past.

I see you are working with a somewhat older version of DSS 7.0 documentation.  Is that the version of Dataiku's Data Science Studio (DSS) you have installed?  The current version of DSS is Version 9.0.1. The same documentation material for the current version of DSS is here.  That said it is very similar to what you have.  That said it might be helpful to other community members if you would be willing to share the version of DSS you are actually using.

I see some comments on the forums that suggest that in the past, that problems that include "with_target" in errors are associated with the version DSS and the need to re-run models in the current version of DSS.  However, this does not seem to apply to you based on your description.

It might also be helpful to provide some of the error messages from the Job Details.

--Tom
0 Kudos
jianhao18
Level 1
Author

Hi @tgb417, thanks for the response.

Yes I am using DSS 7.0.3. I have also shared the error messages from the job details below.

tmp2.jpg

0 Kudos
tgb417

@jianhao18 ,

I donโ€™t know the answer.  I would suggest putting in a support ticket.  Maybe someone else can jump in.  

--Tom
0 Kudos