The Altair Community is migrating to a new platform to provide a better experience for you. In preparation for the migration, the Altair Community is on read-only mode from October 28 - November 6, 2024. Technical support via cases will continue to work as is. For any urgent requests from Students/Faculty members, please submit the form linked here

Changing the positive class when showing f-measure performance measure

bashabora95bashabora95 Member Posts: 3 Learner III
edited November 2019 in Help

Hello dear RapidMiners! 

I have been using RapidMiner for the last couple of months and lately, I am doing a Project with it. 
I have a binominal class label with yes and no. The dataset is unbalanced so I need to upsample it (by using the Filter Examples and Append operators). The problem is that my class of interest is the yes, but that is the minority class. Somehow when I apply the algorithms and also the binominal classification Performance operator, when I see the f measure confusion matrix it has chosen no as the positive class. 
I tried everything, like sorting the values of the attribute or remapping but none made a change (i also tried them in different positions of the process). The only thing that made the trick was to do the connections between Filter Examples and Append a bit differently. Instead of connecting them like : 1st 'exa' port of 'Filter Examples' with 1st 'exa' port of 'Append', and the ori port of 'Filter Examples' with the 2nd exa port of 'Append', I connected them in a crossed way: the ori port of 'Filter Examples' with the 1st exa port of 'Append' ,and the 1st exa port of 'Filter Examples' with 2nd exa port of 'Append'. 
I wanted to ask if this is the correct way to solve this issue, or I should not do it because it doesnt make sense, or just continue with it because it doesnt make any great difference in the logic of how it works? (im not much familiar with the difference between exa and ori) 

Thank you very much in advance,
Bora.

Tagged:
Sign In or Register to comment.