23,303 reputation
13191
bio website biostat.mc.vanderbilt.edu/…
location Nashville, TN
age
visits member for 3 years, 8 months
seen 9 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.


20h
comment Forcing the order of variable in a decision tree in R
In addition, before you get very far test that CART actually works for your sample size (which I suppose is huge) by running CART 10 times on bootstrap resamples of your dataset and checking that the trees produced are very similar.
1d
revised Fractional polynomials vs GAMs
used a more appropriate word
1d
revised Fractional polynomials vs GAMs
edited tags
1d
answered Fractional polynomials vs GAMs
2d
comment How to evaluate the optimal cutoff of ROC curve related to logistic regression using roc from the R package pROC?
Here's a semi-good analogy. A track coach wants to assess an athlete's sprinting speed. She times the a sprint. Does she record "fast" or "slow"? Optimum decisions are made using full information (actual speed; probability of the truth of a statement or probability of an event).
2d
comment How to evaluate the optimal cutoff of ROC curve related to logistic regression using roc from the R package pROC?
Please explain what your utility/loss/cost function is and why you are seeking a cutoff. The end product of logistic regression is intended to be a risk estimate.
Dec
15
comment The impact of non-normality (contaminated normal) on type 1 error and power when sigma is known
A subtle problem: when the data are non-Gaussian the standard deviation may not be very relevant.
Dec
15
comment What is the difference between test set and validation set?
No, unless the dataset is huge or the signal:noise ratio is high. Cross-validation is not as precise as the bootstrap in my experience, and it does not use the whole sample size. In many cases you have to repeat cross-validation 50-100 times to achieve adequate precision. But in your datasets have > 20,000 subjects, simple approaches such as split-sample validation are often OK.
Dec
15
answered What is the difference between test set and validation set?
Dec
13
comment How binary quantile regression divides the dependent variable into quantiles
I looked at those references and don't think they are useful for this problem. The first reference doesn't even use the word "quantile".
Dec
13
answered How binary quantile regression divides the dependent variable into quantiles
Dec
13
revised Ordinary least squares regression giving wrong prediction
Corrected misleading title
Dec
10
comment LogisticRegression - binary classification, “custom threshold”
Not exactly. There is value if having a gray zone with 'no decision'. Utility functions have no notion of 'most probable'. The utility function can be changed as often as someone changes their utilities, even though you don't change your interpretation of the predicted probability.
Dec
10
answered LogisticRegression - binary classification, “custom threshold”
Dec
10
revised How to make prediction in survival analysis using R?
fixed error in function name
Dec
10
comment How to make prediction in survival analysis using R?
You requested a model fit then stated the exponential formula in your computation. How much more clear can I make this?
Dec
10
revised How to make prediction in survival analysis using R?
added 140 characters in body
Dec
10
answered How to make prediction in survival analysis using R?
Dec
10
comment Modelling Technique
Consider a Heckman two-stage model, and note that classification accuracy is not a good accuracy measure.
Dec
10
comment Describing Results from Logistic Regression with Restricted Cubic Splines Using rms in R
Yes. I recommend predictive mean matching. You can look at the R Hmisc package aregImpute function. There is no amount of missing data that would make it better to drop variables or observations entirely. The indicator variable method fails even with a small fraction of missings. Of course your model will be limited when missingness is high, e.g., high standard errors and more dependence on assumptions.