none
Logistic Regression - Where can I see the constant coefficient (biased term) and the Standard Error of the Cofficient?? RRS feed

  • Question

  • Hello together,

    I built a few Logistic Regression models. To analyze them deepr it would be great to know the constant coefficient and the Standard error of the coefficient.

    Where can I see them?

    Thanks and regards

    Basti

    Saturday, June 11, 2016 11:27 PM

Answers

All replies

  • Hi Basti,

    I just built a quick multiclass classification experiment with the Multiclass Logistic Regression algo, and got the following from visualizing the output port of the Train Model module.  I suspect it isn't everything you're looking for though (but please confirm.)

    Thanks,

    jmp

    jmp

    Monday, June 13, 2016 12:42 PM
    Moderator
  • Thanks for the reply.

    I haven't recognized that the bias term is in the Training set because I have so many variables due to dummy variables and the bias term is stated somewhere in the middle. So thanks for that hint!

    Because of the many variables, I have the feeling that I don't see all parameters. I wanted to Export the values by adding a convert to CSV module after the Training set but this is not possible. Is there any way to Access all Parameters from the Training set?

    Still open is the question with the Standard Error Coefficient - I can't find it. It is a measure of certainty found by the logistic regression. The lower this value the lower is the standards error. Moreover, allows this measure to calculate (in Connection with the bias term) the Confidence Interval.

    Thanks and regards,

    Basti

    Thursday, June 16, 2016 10:17 PM
  • Basti and I are working this via private discussion - will post solution when we have it.
    Monday, June 20, 2016 2:05 PM
    Moderator
  • Basti, this is being looked at and considered as an improvement to the product. We will keep you posted as the plans progress.
    Friday, June 24, 2016 2:44 AM
    Moderator