Is this a correct way of proof that logit is linear?

30 Views Asked by At

I am trying to proof that logit is a linear model using the asumption that logisitic regresion is to a linear model?

We start from that asumption:

$$P(Y=1|X_1,\cdots X_p) = \frac{1}{1+e^{-(\beta_0+\beta_1X_1+ \cdots +\beta_pX_p)}} = \frac{1}{1+e^{-z}}$$

Proof

We can define logit like:

$$logit = log(\frac{p(X)}{1-p(X)})$$

Usin the first asumption $$\frac{1}{1+e^{-z}}$$

We have:

$$logit = log(\frac{\frac{1}{1+e^{-z}}}{1-\frac{1}{1+e^{-z}}}) = log(\frac{\frac{1}{1+e^{-z}}}{\frac{1+e^{-z}-1}{1+e^{-z}}}) = log(\frac{1}{e^{-Z}}) = log(e^{z}) = Z = \beta_0+\beta_1X_1+ \cdots +\beta_pX_p$$

So logit is linear

Is this a correct way, or how can i proof that using the first asumption?

Thanks