solution

A storm on July 4, 1999 with wind speeds exceeding 90 miles per hour hit the Boundary Waters Canoe Area Wilderness (BWCAW) in northeastern Minnesota, causing serious damage to the forest. A study of the effects of the storm surveyed the area and counted over 3600 trees to determine whether each of them was dead or alive (data blowdown from package alr3). One of the objectives of the study is to learn the dependence of survival on species, size of the tree, and on local severity. The data set includes results from 3666 trees, including whether a tree was dead or alive (y=1 or y=0), its diameter (D in cm), local severity (S proportion of trees killed), and species (SPP: BF= balsam fir, BS= black spruce, C= cedar, JP= jackpine, PB= paper birch, RP= red pine, RM= red maple, BA = black ash, A= aspen). Fit a logistic regression model and discuss the dependence of survival on the three potential predictors (size, local severity, and species).