Skip to main content
dalguacil1-Visitor
1-Visitor
June 26, 2017
Solved

wrong result interpretation in beanpro example

  • June 26, 2017
  • 4 replies
  • 3085 views

Hello everyone,

I have followed the tutorial and I have got the results for a predictive scoring over the scoring data subset. However, I have only got 8 printhead error over the 150 existing.

I have tried a simple linear regression and the result is worst, because the prediction result is zero error.

I don't know which parameter I must tune to get the boolean result. I suspect that it can be caused by some threshold to identify the result as one or zero.

Could you help me with the model configuration?

Thank you very much.

Best answer by cmorfin

Hi David

One additional point of interest.

The demo json file define the PrinterHead_Error_Occurrence field as DOUBLE.

You can change the json file to define it as BOOLEAN.

The scoring result will be directly 0 or 1.

Regards

Chrisotphe

4 replies

cmorfin19-Tanzanite
19-Tanzanite
June 26, 2017

Hi David

Could you please post the following:

- output of http://<AnalyticsServer>/1.0/about/versioninfo

- the scoring result csv file for the initial model created with the 4 default learners.

Thank you

Kind regards

Christophe

cmorfin19-TanzaniteAnswer
19-Tanzanite
June 26, 2017

Hi David

One additional point of interest.

The demo json file define the PrinterHead_Error_Occurrence field as DOUBLE.

You can change the json file to define it as BOOLEAN.

The scoring result will be directly 0 or 1.

Regards

Chrisotphe

dalguacil1-VisitorAuthor
1-Visitor
June 27, 2017

Thank you very much. Now it is working as I expected

Terms & ConditionsAccessibility statement