About Rick_SAS

Rick_SAS · ‎01-23-2012

You don't specify the distribution, but I assume you want normally distributed values from N(3,1)? Here is a PROC IML solution: proc iml; x = j(20, 30); /* allocate 20x30 matrix */ call randseed(123); /* seed random seed */ call randgen(x, "Normal", 3, 1); /* N(3,1) */ For more about generating random samples, see http://blogs.sas.com/content/iml/2010/09/22/efficient-sampling/ If you one day want a sample from a multivariate normal distribution, see http://blogs.sas.com/content/iml/2011/01/12/sampling-from-the-multivariate-normal-distribution/

Rick_SAS · ‎01-20-2012

I think you can use the ideas in this blog post: http://blogs.sas.com/content/iml/2011/03/16/sorting-a-matrix-by-row-or-column-statistics/ The post uses the SORTNDX function, which is explained further in this post: http://blogs.sas.com/content/iml/2011/03/14/sorting-rows-of-a-matrix/ In your example, IDX is not a valid index into X because IDX contains the value 10. The expression x[10] does not make sense because X has fewer than 10 elements. If you define r=rank(idx), then zz=x[r,] is defined. Ranks and sorts can be confusing because sometimes you need to use the rank and sometimes you need to use the anti-rank.

Rick_SAS · ‎01-19-2012

For earlier releases of SAS/IML, you can use the following module, which appears in my book Statistical Programming with SAS/IML Software: /* Var: return sample variance of each column of a data matrix */ /* x is a matrix that contains the data */ start Var(x); mean = x[:,]; countn = j(1, ncol(x)); /* allocate vector for counts */ do i = 1 to ncol(x); /* count nonmissing values */ countn = sum(x[,i]^=.);/* in each column */ end; var = (x-mean)[##,] / (countn-1); return ( var ); finish; For STD, use sqrt(var(x)).

Rick_SAS · ‎01-17-2012

Use the BY statement: proc sgplot data=SCORE; BY ID; series x=time y=SCORE1; series x=time y=SCORE2/y2axis; xaxis label='time'; run; If you ever want to do this when the ID variable is not already sorted, you will need to call PROC SORT first to sort the data by the ID variable. If this isn't what you want, you can also use the WHERE statement as the first line of the PROC. For example, WHERE ID=1;

Rick_SAS · ‎01-17-2012

In 9.2, I believe you can use the VECTOR statement in PROC SGPLOT to accomplish this. See http://blogs.sas.com/content/iml/2011/06/29/add-a-diagonal-line-to-a-scatter-plot/ Instead of using a macro variable for x0 and y0 as I did in my blog, make these variables in the data set.

Rick_SAS · ‎01-12-2012

Why not use PROC BOXPLOT with ODS graphics? http://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_boxplot_sect031.htm

Rick_SAS · ‎01-12-2012

One general bit of advice: it is usually a poor idea to interlace macro code and IML. It is almost always unnecessary, and it makes debugging a real pain. By restructuring the logic, you can usually avoid %IF/%THEN and other macro logic and use IML statements instead.

Rick_SAS · ‎01-10-2012

There is no built-in pre-written algorithm to do this. The following web page outlines an implementation in MATLAB or SAS/IML: http://www.angelfire.com/falcon/isinotes/statcomp/mcmc/gibbs5.html

Rick_SAS · ‎01-10-2012

In the example that you provide, the arsumc matrix is a 30x30 matrix with most elements about 1E130. That's the source of the error reported by GINV. if iteration=1 then do; _min = min(arsumc); _max = max(arsumc); _mean = arsumc[:]; print _min _max _mean; end; _min=6113822.2 _max=1.589E133 _mean=2.959E131

Rick_SAS · ‎01-10-2012

For 1D, use the CDF function. For 2D, use the PROBBNRM function. For higher dimensions, Genz's 1992 algorithm was implemented in IML by Bretz: http://www.biostat.uni-hannover.de/89.html Somerville and Bretz also have a 2001 algorithm that is available on the same web site. I haven't used the high-D algorithms, so I can't speak to their efficiency. Rick

Rick_SAS · ‎01-09-2012

The neqative binomial (NB) model should have a parameter, k, so I don't think your formula is correct. Your first question is statistical, rather than having to do with matrices and linear algebra. I'm not an expert in GEEs or using generalized linear models, but I discussed this with a colleague. As best we can tell, the macro writer is using the following definitions for a model with link function g(): Ui: mean Fui: ginv(mean), i.e. linear predictor Fui_dev: diagonal matrix of weights for Fisher scoring = 1/(v(mu)*dg(mu)**2) (not sure about this one: gamma has negative sign?) Vui: diagonal matrix of inverse of square root of variance. If these are right, for a log-linked NB with dispersion parameter k, you might try Ui = xi*beta Fui = log(ui) Fui_dev = diag(ui/(1+k*ui)) Vui = diag( 1/sqrt(ui+k*ui##2)) As I've said, this is a guess. You might get better answers from the SAS Discussion Forum on SAS/STAT and Statistical Procedures. For your second question, I'm away from my office, so can't reproduce the error. Perhaps when you correct specify the NB model, this second error will go away. Or someone else might be able to help.

Rick_SAS · ‎01-06-2012

Without additional details, it will be hard to help. Have you tried a different initial condition? The initial step is determined by the initial guess.

Rick_SAS · ‎01-06-2012

Sort by year.Then use YEAR as a BY variable and use the OUT= option to save the kernel densities to an output data set. You can use the SERIES statement in PROC SGPLOT to overlay the curves. If you don't want them overlaid, but are content with a panel, you can use PROC UNIVARIATE. In this case, use YEAR as a CLASS statement. See this example: http://support.sas.com/documentation/cdl/en/procstat/63963/HTML/default/viewer.htm#procstat_univariate_sect070.htm

Rick_SAS · ‎01-06-2012

Welcome! Thanks for a great question! It sounds like you have variables such as Y2000-Y2011 which represent different yearly measurements of some quantity. You can use the PLOTS=densityoverlay option to overlay the densities of all of these onto a single plot. Here is an example of overlaying two densities in a SASHELP data set: /* use KDE to overlay estimates of different variables */ proc kde data=sashelp.cars; univar mpg_city mpg_highway / plots=densityoverlay; run;

Rick_SAS · ‎01-03-2012

Wow. Answers like this give me a warm feeling inside. On behalf of all of us who don't have as much experience, thanks to LVM and Steve and ALL of the SAS experts who spend considerable time to answer these technical forum questions!

Online Status	Offline
Date Last Visited	Friday