PhD student.

Jul
2
awarded Curious
May
27
awarded Tumbleweed
May
20
asked Poisson regression on partially missing covariates
May
11
comment Can I trust a model if I do not check the prediction error?
Related to Glen_b's comment: darden.virginia.edu/web/uploadedFiles/Darden/Faculty_Research/…
Apr
25
comment Normalization of dummy variables
You might find this paper by A. Gelman interesting stat.columbia.edu/~gelman/research/published/standardizing7.pdf
Mar
12
comment Simulate predicted probabilities after a logistic regression
See here: stata.com/support/faqs/statistics/…
Feb
16
awarded Yearling
Feb
16
awarded Yearling
Jan
2
awarded Nice Answer
Nov
11
awarded Commentator
Nov
11
comment Why do so many people apply to so few PhD programs?
PhD students are not chosen randomly from those who apply!
Oct
10
comment SNP genotype coding in regression
It should be the other way around. If you're assuming a gene-dosage effect you have only one parameter and it's a one degree of freedom test. If you dpn't assume the gene-dosage effect, you have 2 parameters and if you want to test them jointly it's a 2 dof test.
Oct
5
comment Concerns about the size of odds-ratio estimates in binary logistic regression model
$\exp(0)=1$ and not $0$ (in the last row, [L1 Reading=0])
Oct
2
awarded Yearling
Sep
30
comment Difference between Maximum a posteriori and penalized likelihood
I don't know if this can be of any help, but - for example - quadratic loglikelihood penalization (squared $L_2$ norm) corresponds to having independent normal priors on the model parameters. I don't know the deatils of the paper you're referring to, but it's definitely not impossible to "exploit" PL for that purpose - Or maybe I've misunderstood your question?
Sep
29
comment What statistical test is appropriate when comparing heart rates from different devices?
Since apparently you (@phedonrousou) have repeated measures on the same subjects, this paper might prove useful: Myles & Cui, Using the Bland–Altman method to measure agreement with repeated measures - British Journal of Anaesthesia 309–11 (2007), doi:10.1093/bja/aem214
Sep
26
comment Multiple imputation of time variables -- which step to impute?
Before turning that comment into an answer... I was thinking, t is likely to be a censored r.v., right? I think this is a major issue when it comes to imputing t using, for example, MICE. But what about quantile imputation? Look for example at this: stata.com/meeting/sweden11/abstracts/bottai_nordic11.pdf if you don't have administrative truncation of the follow-up, this might be a solution. Imputation of missing values using laplace regression can take care of the fact that t is (most likely very skewed and) censored. @EpiGrad
Sep
25
comment Multicollinearity and Splines :: Is there a problem?
You can always orthogonalize the generated splines (for example the rcsgen Stata command uses Gram-Schmidt orthogonalizaton)
Sep
23
awarded Constituent
Sep
17
awarded Caucus
1 2 3 4 5