I'm working from a project built by a previous programmer. The call to glm() the programmer has in their documentation is
glm(formula = AVAL ~ AUC, family = binomial(), data = logreg.dat)
I have the output from that code, but none of the data used to build the model.
Deviance Residuals:
Min 1Q Median 3Q Max
-0.5991 -0.4784 -0.4220 -0.3028 2.7425
Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -1.62668 0.34738 -4.683 2.83e-06 ***
AUC -0.14730 0.07448 -1.978 0.048 *
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
(Dispersion parameter for binomial family taken to be 1)
Null deviance: 119.87 on 193 degrees of freedom
Residual deviance: 113.23 on 192 degrees of freedom
AIC: 117.23
Number of Fisher Scoring iterations: 6
Is there a way to rebuild this fit? I'd like to use the fit to simulate out some new AVAL data based on a new AUC vector.
If I correctly understand, you want to simulate the model using the estimates of the fit for the parameters. You can do:
# take the estimates from the output
intercept <- -1.62668
slope <- -0.14730
# new AUC values
AUC <- c(1, 2, 3, 4, 5)
# simulate AVAL values
logit_probs <- intercept + slope * AUC
probs <- 1 / (1 + exp(-logit_probs))
n <- length(AUC)
rbinom(n, 1, probs)