Scorer for the sum of a column

Hello,

Scorer node is helpul to find the max number of successful predictions.

Is there a node (or workflow example?) to do a score based on the best sum of a column instead of the best number of successful or failed prediction ? 

 

Thank you

Nobody could help me ?

Hi Lug,

could you make us an example what you want to achieve? What do you mean with the best sum of a column?

 

Maybe you are looking for the numerical scorer? Or something like the groupby node?

Best regards, Iris 

Hello,

For example, I have a table like the follwing one but with many lines and columns:

Car Sexe Age Job City Benefit
Car1 F 25 Job1 City1 880
Car2 M 65 Job2 City1 -500
Car3 M 42 Job3 City2 680
Car4 M 45 Job2 City3 -1500
Car5 F 34 Job1 City4 -700
Car6 M 28 Job5 City5 280
Car7 M 51 Job4 City2 1640
Car2 F 42 Job3 City5 -450
Car6 F 38 Job1 City3 460
Car2 M 32 Job5 City2 -300

I use a predictive model with Random Tree Forest to predicy with which conditions I will have benefits. It is based on the example 04_Analytics/11_Optitimization/06_Parameter_Optimization_Two_Examples. But instead of measuring 

The Scorer node gives me the ratio between good predictions and bad predictions (number of rows with positive benefit vs number of rows with negative benefit). Would it be possible to score the result on the benefit column sum (resulting of the prediction) to find the best option for my model ?

Thank you

Hi Lug,

Maybe instead of using Scorer node you could try GroupBy node. Leave Groups Tab empty and in Manual Aggregation Tab calculate the Sum on the Prediction column. I made a small workflow to illustrate.

Hope it helps.

Best,

Daria