-1
$\begingroup$

When I run a logistic regression using sm.Logit (from the statsmodel library), part of the result looks like this:

enter image description here

The variable y is academic success (average greater than 10). The problem is that I am lost in front of this table (there is only the beginning), I do not know how to interpret the results. In the coef column, can I interpret only the sign or the value too? For "absences" I can say what with -0.0459? Also what does column "z" mean and what can I say on it? The confidence interval I understood but the beginning is quite blurry for me, can you help me please?

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ If this is logistic regression then the variable is presumably the log-odds of academic success, so if the probability of success is $p_i$ then this is estimating a regression for $\log\left(\frac{p_i}{1-p_i}\right)$ in the form $\beta_0+\beta_1\text{ studymore} + \beta_2\text{ absences} + \beta_3\text{ failures} +\cdots$, and you can use this to transform the estimated log-odds back to probabilities $\hat p_i$. The estimated coefs are the $\beta$s with their std err (standard errors), and the confidence intervals are close to these coefficients $\pm \, 2$ standard errors. $\endgroup$
    – Henry
    Commented May 24, 2023 at 12:17

1 Answer 1

Reset to default
3
$\begingroup$

The logistic regression model is

$$ E[y|X_1,X_2,\dots,X_k] = \sigma(\beta_0 + \beta_1 X_1 + \beta_2 X_2 + \dots + \beta_k X_k) $$

The coef column displays the $\beta_0,\beta_1,\dots$ parameters, where $\beta_0$ is described as const. On the interpretation of the parameters you can read in the following threads:

The z column shows the $Z$ values discussed in:

Those are just examples of multiple threads we have, so check other questions tagged as for more details.

Improve this answer
$\endgroup$
1
  • $\begingroup$ I'm sorry but I don't really understand, with your second link if I understand correctly, the effect of "absences" corresponds to (1-0.9545) = 0.0455 so a 4.55% decrease in the chance of having the average ? it seems strange to me. I need to do this regression for a little homework, if I just interpret the significance of the coefficients with their signs, can this be enough as an analysis? (for significance I look at P>|z|?) $\endgroup$
    – mferrer47
    Commented May 24, 2023 at 12:53

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