For the sake of future readers who come to this thread, you can read about three different methods for finding the largest (or smallest) values in each group: Display the largest values for each group - The DO Loop (sas.com)   The DATA step with BY-group processing  PROC RANK PROC MEANS with the IDGROUP option on the OUTPUT statement ... View more

> Is it correct to say that sometimes quantile regression gives the same results, but occasionally it might differ? Yes. And as balardw points out, the estimates also depend on the univariate method used to estimate the quantiles. As with all statistics, there are many ways to form an estimate from the data.   ... View more

> The median for one group is 72, and for the other group, it is 40. Therefore, the expected median difference should be 72 - 40 = 32.   No, that's not how quantile regression works. You do not necessarily get the same estimates as you would for computing the median for each group separately and then subtracting. For example, the following code uses PROC MEANS to compute the median MPG_City value separately for Origin=Asia and Origin=USA. The results are 20.5 for Origin=Asia and 18 for Origin=USA. Thus, a naive estimate of the difference is 2.5. The program then uses your code to perform a quantile regression. The QUANTREG estimates are 20 and 18 for a difference of 2.0. data cars; set sashelp.cars; if Origin in ('Asia' 'USA'); run; proc means data=cars median; class Origin; var MPG_City; run; proc quantreg data=cars ci=resampling; class Origin; model MPG_City = Origin / quantile=0.5; estimate 'Diff in Medians' Origin 1 -1 / CL; run; For a discussion of how QUANTREG differs from the naive estimate, see "Quantile regression: Better than connecting the sample quantiles of binned data."  For an additional example of how the QUANTREG estimates can differ from univariate estimates, see "Quantile estimates and the difference of medians in SAS."  ... View more

I suggest you start a new thread with your question, example data, and a description or picture of what you want to display. ... View more

This can happen with data that are rounded. For a discussion, example, and solution, see  The mystery of the density curve that was too short - The DO Loop (sas.com) To make this visual illusion disappear, use a bin width that is at least as large as the rounding unit in the data.  ... View more

Check the definition of the CENSOR variable in your data? Is it possible that you should be using CENSOR(0) instead of CENSOR(1) on the TIME statement? If you accidentally switch the 0/1 value, you can get the situation that you describe.   ... View more

How about this?   proc sort data=have2; by CustID Var; run; proc rank data=have2 out=want(where=(rank <= 3)) /* keep top 3 ranks */ ties=low; by CustID Var; var value; ranks Rank; run; proc sort data=want; by CustID Var Rank; run; proc print data=want;run; ... View more

Ultimately, the editor and reviewers will help you decide if it is "okay" to show both graphs. Be sure to explain the model that each graph visualizes. Since they are different models, it shouldn't be surprising that the graphs are different. ... View more

For future reference, here are few related articles: Moment-ratio diagram and finding the (a,b) values so that Beta(a,b) has a specified skewness and kurtosis Use the moment-ratio diagram to visualize the sampling distribution of skewness and kurtosis Bimodal and unimodal beta distributions ... View more

I think so. Can you show the code that you are running to fit the Cox model? ... View more

It's not PROC SGPLOT that is responsible for the difference in the predicted values, it is your regression model. The second model has an addition 12 observations before (and 3 after) the original observations. These observations cause the regression to choose a different model for the second set of observations. (Compare the parameter estimates for both models!) The graph is simply showing you the model in both cases, and they are different models. ... View more

1. You don't need to transpose data when you are taking the MAX. So all statements that look like max(vector`) can be rewritten as max(vector) 2. In fact, I don't know why you are transposing any of the vectors. You can use column vectors instead of row vectors and save 6*ncol(b) vector transposes per iteration. 3. It looks like sess_usage_upto is only used as the ratio sess_usage_upto/pos. Therefore, you can compute the ratio one time by modifying the 3rd column: x[,3] = x[,3] / x[,2]; Then there is not need for the ratio inside the LOGISTIC function argumet. 4. Regarding the LOGISTIC call, if you stop transposing and use column vectors, you can use matrix multiplication inside the LOGISTIC call. The linear combination of the vectors is equal to x[idx,2:4)*param[2:4] or x[idx,2:4)*param[2:4]`  (not sure which)   These tips might give you another 5-10% improvement, but the size of the data set is what is hurting the performance. ... View more

You might try to use a decision tree instead of logistic regression. Logistic regression drops the entire observation if any variable is missing. A decision tree doesn't. See the PROC HPSPLIT documentation at https://documentation.sas.com/doc/en/statug/15.2/statug_hpsplit_examples01.htm   ... View more