search for: shuangge

...with the following glm question? for the R function glm, what I understand is: once you specify the "family", then the link function is fixed. My question is: is it possible I use, for example, "log" link function, but the estimation approach for the guassian family? Thanks, Shuangge Ma, Ph.D. ******************************************** * CHSCC, Department of Biostatistics * * University of Washington * * Building 29, Suite 310, 6200 NE 74th ST. * * Seattle, WA 98115 * * Tel: 206-685-7123 Fax: 206-616-4075 *

Dear Sir/Madam: this is shuangge Ma, graduate student in UW-Madison statistics department. I have a computational question. I have a function f(x,y). I want to find the y(x) that maximize f(x,y) under the constraint y(x) is a non-decreasing step function. Is there any R package or algorithm I can use for this purpose? thanks a lo...

...f is a known function an X_{i} are i.i.d. r.v. I need to maximize the above sum, under the constaint that: beta'X_{i}+alpha<=1, for i=1,...,n. For one dimension, it is kind of trivial. What should I do with high dimensional alpha and beta? Thanks for your time, Shuangge Ma, Ph.D.

...he return value "convergence" of the function "optim", the values should be 0, 1, 10, 51 and 52. See http://www.maths.lth.se/help/R/.R/library/stats/html/optim.html When I use constrOptim, I get "convergence" values 7 and 11. What do they mean exactly? Thanks again, Shuangge Ma, Ph.D.

...I have a question about spline: ns (or bs) in R (library: splines). What's the exact analytic form of the B splines generated by ns (or bs)? I need to calculate the penalty (square of second order derivative) of the basis functions. Is there any shortcut or I need to program myself? thanks, Shuangge Ma ********************* Department of Statistics Phone: 608-263-4782(O) University of Wisconsin--Madison

Hello, I will really appreciate some help with the following question. let's say cor.m is a correlation coefficient matrix. tree<-hclust(as.dist(1-cor.m)); plot(tree); my question is: is it possible to add the correlation coefficient information to the plot? thanks, S. Ma

hello, I have the following questions with the R package lars(). I would really appreciate som help here. 1. do I have to standardize the predictors and the response? 2. the function cv.lars(): how can I specify it is for lasso not stagewise or lars? your help will be really appreciated. thanks,