**Harvard, HST 951**

**Excerpt:** ... a function of the low birth weight indicator), as well as a correlation matrix. What results appear the most striking to you? Create a random test set of 200 cases and show whether your sample is representative of the population. If not, try another one. From here on, calculate the Brier score, the c-index, plot the ROC, and calculate the Hosmer-Lemeshow (HL) goodness-of-fit (both types) for all requested models in the training and test sets. Use R to run your ** logistic regression model **s, and either netlab or NevProp from http:/brain.unr.edu/FILES_PHP/show_papers.php#software. Do not use off-the-shelf ROC software please! (ii) Run a ** logistic regression model ** using maternal smoking to predict low birth weight. What are your inferences? What is the estimated probability of having a low birth weight infant for a woman who smokes 30 cigarettes per day? What is the estimated probability of having a low birth weight infant for a nonsmoking woman? What is the estimated odds ratio associated with a 30 cigarette per ...