Producing binary classification [Yes (1),No (0)] instead of scoring (with Score Tool)

[muhalim]

Hi, is it possible to get the prediction results as Yes/No (1/0) instead of scoring as produced using Score Tool ? Thank you in advance.

[fmvizcaino]

Hi @muhalim ,

What you can do is to insert a formula tool with an if clause separating the score by half.

If score < 0.5 then 0 else 1 endif

Best,

Fernando V.

[muhalim]

Thank you. 

Is it:

     if [Score_N] < 0.5 then 0

  else 1

  endif

[DiganP]

@muhalim You can use the formula tool to create the cut off point. I don't think you should use 0.5 as the cutoff point however. The regression tools create an optimal cutoff point in the Interactive output, which you can use in the formula tool conditional statement. 

ex:

if [score]>0.23 then 'yes' else 'no' endif

Here, 0.23 is the optimal cutoff output from the model. 

[fmvizcaino · Accepted solution]

Hi @muhalim ,

It is Score_Y>0.5, but it also depends on your evaluation of the confusion matrix where you can analyze what you want to look for.

Here you can learn a little about how to define your threshold.

https://towardsdatascience.com/understanding-confusion-matrix-a9ad42dcfd62

Best,

Fernando Vizcaino

[muhalim]

thank you 

[muhalim · Accepted solution]

thank you