Scorer Accuracy Statistics Calculations

Hi,

I'd like to post-process the scorer accuracy statistics table because I have multiple iterations and would like to find the best random seed or iteration to use.  I attached a file that shows pre/post proccessed and desired data.  I'm not familiar with Java but is this something that requires it?

Thanks.

Hi, 

I'm guessing a bit at what you are looking for exactly, but is it something like the flow attached?  Have a look and post back here if you will. 

Regards,

Aaron Hart

KNIME.com

 

 

Thanks! What really helped was the statistics node because it calculated the mean average of the scorer's accuracy statistics. Then, as suggested, I used row filter to include only the means. 

One more question (I hope) that's related to this workflow: How do I pass a cell's value that holds a specific random seed INTO a partitioning node's random seed?

My pleasure.  To pass a seed (or any setting) downstream in a flow you just need to assign it a variable name using the flow variables tab.  Subsequently you will be able to reference this variable using the specified name.

A simple example is attached.

Regards,

Aaron

Aaron, actually I'm trying to pass a cell's value into the next node's random seed.  So in the attached file I'd like to pass the top-right value 918245 as a flow variable into a partitioning node's random seed. 

Look to the tablerow to variable and/or the tablerow to variable loop start nodes.  They will help you convert the cell value to a flow variable for use in downstream processing.

Regards,

Aaron

Works like a charm, thanks!