Combining non-parametric models with logistic regression: an application to motor vehicle injury data

摘要: To date, computer-intensive non-parametric modelling procedures such as classification and regression trees (CART) multivariate adaptive splines (MARS) have rarely been used in the analysis of epidemiological studies. Most published studies focus on techniques logistic to summarise their results simply form odds ratios. However flexible, CART MARS can provide more informative attractive models whose individual components be displayed graphically. An application these sophisticated an case-control study injuries resulting from motor vehicle accidents has encouraging. They not only identified potential areas risk largely governed by age number years driving experience but also identify outlier groups a precursor detailed analysis.