Using Log to label results

rakirk

I have a process that basically involves 7 different learners and then calculates a t-test for significant differences in accuracy across these models. The problem is that I am having trouble figuring out which element is which in the resultant t-test table. I have tried using the log to record the accuracy of the models and then compare afterwards, but I'd like to have label creation be a part of the process. Any ideas?

rk

land

Hi,
you can use the Reporting Extension to give you static reports with sections and subsections. You can also rename the performance operators and t-tests. The name of the operator where they are generate is shown for each object in the result view.

Greetings,
Sebastian

rakirk

Hi Sebastian,

Thanks for sharing your expertise. I keep coming back to this hobby project and lately I've been trying to find a way to report the level of the false negative and false positives within a classification scheme. I would, ideally, find a way to get these results using the `log` operator; I currently use `log` to report accuracy levels within the `performance` operator. It is not feasible to extract these from the results pane since there are several loops created that result in over 100,000 iterations. I am, basically, trying to do some meta-mining of classifier accuracy. Let me know what you think,


rk