24,763 reputation
13597
bio website biostat.mc.vanderbilt.edu/…
location Nashville, TN
age
visits member for 3 years, 9 months
seen 10 hours ago

I am Professor of Biostatistics and Chairman of the Department of Biostatistics at Vanderbilt University School of Medicine, Nashville TN USA. I am Associate Editor of Statistics in Medicine, a member of the Faculty of 1000 Medicine, and a member of the policy advisory board for the Journal of Clinical Epidemiology. I am a Fellow of the American Statistical Association. I am author of Regression Modeling Strategies (Springer, 2001). My specialties are development and validation of predictive models, clinical trials, observational clinical research, cardiovascular research, technology evaluation, clinical epidemiology, medical diagnostic accuracy, biomarker research, pharmaceutical safety, Bayesian methods, quantifying predictive accuracy, missing data imputation, and statistical graphics and reporting. I am a long-time user of R. In August 2014 I was given the WJ Dixon Award for Excellence in Statistical Consulting by the American Statistical Association. Among many other things, Dr Dixon was the lead developer of the first general-purpose statistical software package, BMD.


1d
awarded  multiple-regression
1d
comment Analyzing residuals in logistic regression
Not in the least. You want to know the probability of getting on a lifeboat as a function of age, class, sex, etc. Just to take an example, to "classify" someone as dead when their probability of surviving is 0.499999 is folly.
1d
answered Analyzing residuals in logistic regression
1d
comment Analyzing residuals in logistic regression
This is a risk prediction problem, not a classification problem, and the indexes you recommended are improper accuracy scores, i.e., they are optimized by choosing the wrong features and giving them the wrong weights.
2d
comment Nonlinear effect in an interaction term
That is fine if there is only one transformation and that transformation is pre-specified. It's not so fine if the transformation is determined by analyzing the data.
2d
comment Nonlinear effect in an interaction term
Multiple attempts at transformation so as to get good residual plots induces a distortion of inferential quantities such as $P$-values, standard errors, and confidence limits.
Jan
27
comment Nonlinear effect in an interaction term
Stata nicely handles restricted cubic splines but you may have to construct the interaction terms yourself, and the anova is not as comprehensive.
Jan
27
revised Nonlinear effect in an interaction term
Fixed model from ols to logistic
Jan
27
comment Nonlinear effect in an interaction term
Whoops. Fixing now.
Jan
27
revised Nonlinear effect in an interaction term
edited tags
Jan
27
answered Nonlinear effect in an interaction term
Jan
24
comment Logistic regression with +1/-1 labels
Kindly do not downvote an answer unless you can show that it is wrong or irrelevant.
Jan
24
comment Logistic regression with +1/-1 labels
The logistic regression model was invented no later than 1958 by DR Cox, long before the field of machine learning existed, and at any rate your problem is low-dimensional.
Jan
24
comment Logistic regression with +1/-1 labels
I'd like to see the general case.
Jan
24
revised Logistic regression with +1/-1 labels
edited tags
Jan
24
answered Logistic regression with +1/-1 labels
Jan
22
revised The best model of an AICc-based model selection on a very small sample has an high number of predictors: does it make any sense?
edited tags
Jan
22
answered The best model of an AICc-based model selection on a very small sample has an high number of predictors: does it make any sense?
Jan
19
comment R code for standardized coefficents and model effect size using nlme
This is not a recommended practice even for ordinary linear models. See elsewhere on the site.
Jan
19
revised Multiple test correction for categorical variables with many levels
edited tags