Getting the function of the Regression in the ExampleSet, Visualisations Register in a Scatter Chart

Carlo

Hi Community,

I would like to get the function of the displayed regression in the visualisations register (Example Set) for a Scatter Chart is there any possibilty?

Thanks a lot for your support.

Carlo

Telcontar120

Instead of deleting the country attribute, you can do this simply by first filtering for the value of each country one at a time, running the regression, and then capturing the associated linear regression model.  You could do this manually, or you could actually automate it easily with a Loop operator in RapidMiner (filter for each country value, build a regression, and save the model).

sgenzer

hi @Carlo - ok a lot going on here. Perhaps you just want to use a Loop Attributes operator? You can then select the subset (countries) to loop, and then use Select Attributes and your Linear Regression inside the Loop Attributes operator. That will likely do the trick.

Scott

Marco_Boeck

Hi,

Currently that is not possible, but for linear regression in Scatter plots the function will be shown in the tooltip with the next patch (9.2.1).

Regards,
Marco

Carlo

Hi Marco,

sounds great - is there already a date defined for the next patch?

Thanks again,

Carlo

Marco_Boeck

Hi,

I cannot give you a definitive date, but right now the patch is planned for the week of the 18th.
Obvious disclaimer: if something goes wrong it may get delayed.

Regards,
Marco

Carlo

Okay thanks

Do I have another option to get the the regression function for several sets of my example set?

Regards,

Carlo

sgenzer

if you just want the regression function you can use the "Linear Regression" operator.

Carlo

Hi sgenzer, thanks for your post. I tried already the Linear Regression operator but my problem is, that I need the Linear regression for a several set of countries. So p.e. I need the Linear Regression for Italy, Spain and Portugal (first set) and second set France, Netherlands, Germany (second set) etc. etc.

The linear regression operator works only with numerical numbers, which means I have to delete the column of the country names, so then I can not set different sets. In the result I have just one regression function for all the data / all the countries...

Telcontar120

Instead of deleting the country attribute, you can do this simply by first filtering for the value of each country one at a time, running the regression, and then capturing the associated linear regression model.  You could do this manually, or you could actually automate it easily with a Loop operator in RapidMiner (filter for each country value, build a regression, and save the model).

sgenzer

hi @Carlo - ok a lot going on here. Perhaps you just want to use a Loop Attributes operator? You can then select the subset (countries) to loop, and then use Select Attributes and your Linear Regression inside the Loop Attributes operator. That will likely do the trick.

Scott