How to interpret 'cost' column in classifications

cramsden

Hi Im new to this whole world and am using rapid miner go. I set up a deep learning model for classification. In the output when I apply my model to a new data set the first column that shows is 'cost'. I assumed that has to do with the 'gain and cost' payoff table that I set up when creating the model, but it doesnt seem to match with what I'd expect to see given the inputs I put in.

Can any one help me on how to interpret this 'cost' column or how it is calculated.

Thank you!

MartinLiebig

Hi,

this should be the lowest cost, where cost is defined as confidence*cost_in_matrix.

Best,

Martin

MartinLiebig

Hi@cramsden,

sure! its the calculation we talk about here://www.turtlecreekpls.com/resource/profit-sensitive-scoring/. maybe there is a minus sign somewhere, because sometimes you talk about costs, and sometimes about gains.

Best,

Martin

cramsden

Yes, I understand setting up the cost benefit table. But I'm asking more for the cost column in the output when I apply the model to new datasets (I'll attach a screen shot). I think I am interpreting this wrong because if I have 2 different models one with the default (+1,-1) payoff table, and another which i adjust to +1,-10000 just to see how that effected the cost column in the output. I expected to see the cost column be large negative numbers with this set up but they were not. That's why I am a bit confused, I cant find any documentation on this.

cramsden

Thank you! Figured it out!!