0
$\begingroup$

I am trying to understand the details of the logistic regression models and now I was wondering how the model can be created if you have the coefficients and intercepts.

So I created a logistic regression model in python and I extracted the coefficients and intercepts. Now I want to calculate those by hand, just to see how it works. Does it make sense to calculate it like a linear regression (y = a + bx1 + bx2 ) and then if the given value is less than your threshold, predict the output as 0, otherwise as 1?

I have already done it but my concern is that some of the predicted probabilities (output) came out negative and a few were more than 1.

Thank you

$\endgroup$
0
$\begingroup$

You first perform the calculation as you would for a linear regression, but before using the treshold to predict either 0 or 1 you first have to pass the result (i.e. the y value from your formula) through the sigmoid function to make sure all values are between 0 and 1. In reality, the formula for logistic regression (before thresholding) would therefore be as follows:

\begin{equation} y = \frac{1}{1+e^{-(\alpha + \beta_1 x_1 + \beta_2 x_2)}} \end{equation}

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.