Adaptive Regression for Modeling Nonlinear Relationships by George J. Knafl, Kai Ding

By George J. Knafl, Kai Ding

This ebook offers tools for investigating even if relationships are linear or nonlinear and for adaptively becoming acceptable versions once they are nonlinear. information analysts will methods to include nonlinearity in a single or extra predictor variables into regression types for various different types of end result variables. Such nonlinear dependence is usually now not thought of in utilized examine, but nonlinear relationships are universal and so have to be addressed. a regular linear research can produce deceptive conclusions, whereas a nonlinear research promises novel insights into information, no longer differently attainable.

A number of examples of the advantages of modeling nonlinear relationships are provided through the ebook. tools are coated utilizing what are referred to as fractional polynomials in keeping with real-valued energy ameliorations of basic predictor variables mixed with version choice according to chance cross-validation. The e-book covers tips on how to formulate and behavior such adaptive fractional polynomial modeling within the typical, logistic, and Poisson regression contexts with non-stop, discrete, and counts results, respectively, both univariate or multivariate. The publication additionally offers a comparability of adaptive modeling to generalized additive modeling (GAM) and a number of adaptive regression splines (MARS) for univariate results.

The authors have created custom-made SAS macros to be used in carrying out adaptive regression modeling. those macros and code for undertaking the analyses mentioned within the e-book can be found throughout the first author's site and on-line through the book’s Springer web site. specified descriptions of ways to take advantage of those macros and interpret their output look in the course of the booklet. those tools might be carried out utilizing different courses.

Show description

Read Online or Download Adaptive Regression for Modeling Nonlinear Relationships PDF

Best biostatistics books

Permutation, Parametric and Bootstrap Tests of Hypotheses: A Practical Guide to Resampling Methods for Testing Hypotheses

Explains the mandatory historical past in checking out speculation and determination concept to allow innumerable functional purposes of records. This ebook comprises many real-world illustrations from biology, company, medical trials, economics, geology, legislations, medication, social technological know-how and engineering in addition to two times the variety of routines.

Transforming Health Care Through Information (Health Informatics)

Reworking future health Care via info, moment variation, attracts upon the adventure of the most important specialists in scientific informatics to deal with the advanced demanding situations confronted whilst healthcare organisations enforce technological adjustments. This book’s case experiences illustrate particular and useful suggestions hired to beat such demanding situations.

Biostatistics

""The booklet offers very important subject matters in biostatistics along examples supplied within the programming languages SAS and R. . .. The e-book covers many proper subject matters each scholar should still comprehend in a manner that it makes it effortless to stick with . .. each one bankruptcy presents workouts encouraging the reader to deepen her/his figuring out.

Extra info for Adaptive Regression for Modeling Nonlinear Relationships

Sample text

These deleted fold likelihoods are normalized by the sample size n and multiplied up to generate the LCV score. LCV scores are geometric averages of deleted fold likelihoods with larger scores indicating better models for the data. Also, all model parameters are used in computing LCV scores, including for regression models the variance parameter σ2 and not just the vector β of parameters for the means as in the LSCV formulation. LOO LCV is the special case of k-fold LCV with k ¼ n folds, each consisting of an index for a single subject.

9). Moreover, fully specified polynomial models can overfit the data, and fully specified polynomial models of higher degree than the cubic model are infrequently used. Royston and Altman (1994) proposed an alternative they called fractional polynomial modeling allowing powers to be real valued. Their work has been extended by Royston and Sauerbrei (2008). 5, 1, 2, 3 and the natural log transform as the 0 power case for degree 1 fractional polynomials with only one power transform. One choice within the fixed set is used, not the complete set as in fully specified standard polynomial models.

Nilson, E. , & Walsh, J. L. (1967). The theory of splines and their applications. New York: Academic Press. Burman, P. (1989). A comparative study of ordinary cross-validation, ν-fold cross-validation and the repeated learning-testing methods. Biometrika, 76, 503–514. Cleveland, W. , Devlin, S. , & Gross, E. (1988). Regression by local fitting. Journal of Econometrics, 37, 87–114. , & Everitt, B. S. (2006). Statistical analysis of medical data using SAS. Boca Raton, FL: Chapman & Hall/CRC. Diggle, P.

Download PDF sample

Rated 4.85 of 5 – based on 28 votes