Obtain the 90% confidence intervals for the logistic regression model chdglm, as discussed in Example 17.5.1. Also, carry out the deviance test to find if the overall fitted model chdglm is a significant model? Validate the assumptions of the logistic regression model.

Data from in Example 17.5.1

Contd. We will now see how the above irls function compares with the R modules. The R function glm will be useful to build GLM, and it works on similar lines as the linear model function lm. The formula ∼ as used in lm continues to hold for glm and does not need further elaboration.

Thus, the irls function returns the precise answer as the R function glm as seen by the output for irls(chdage$CHD,chdage$AGE) and chdglm$coefficients. It is now hoped that the reader is really comfortable with the IRLS algorithm.

The summary function shows that AGE is a significant variable for explaining CHD. Further description will follow in the rest of this section.

Estimates of the link function and odds ratio is straightforwardly obtained by plugging in the values of the estimated regression coefficients, that is,

We will next consider an example of the multiple logistic regression, that is when we have more than one covariates.

Transcribed Image Text:

> data (chdage) > chdglm glm (chdage $CHD~chdage $AGE, family='binomial') irls (chdage $CHD, chdage $AGE)