@Ronein    I think you are looking for something like this:   data want; if _n_ = 1 then do until(eof); set tttbl1 end = eof; sum + Mis; end; fac = (sum - 200) / sum; do until(last); set tttbl1 end = last; new = min(Use, ceil(Mis * fac)); output; end; drop fac sum; run; ... View more

@k93943    In addition to @Reeza 's suggestion for IML, you can do it using only BASE SAS. Proc FCMP provides data step approach to buiding user defined functions.   If I am convinced of your reasons for having such function, I can throw some more light into doing it. ... View more

@fibonacheese    As indicated by @PGStats  try the following:   data poisson; call streaminit (0); retain sumy meantotal; psamples = 20; *2000; psizesamp = 40000; lambda = 8; do psample = 1 to psamples; sumy = 0; do n2 = 0 to psizesamp; y = rand("Poisson", lambda); sumy + y; end; mean = sumy / psizesamp; meantotal + mean; *output; end; mom = meantotal / psamples; put 'Mean of the Means ' mom; run; See the commented statements for lesser / no output. A better way is to keep out the initial values as macro variables for not to load them into PDV as: %let lambda = 8; %let psamples = 2000; %let psizesamp = 40000;   and inserting them into the code. ... View more

@mmohotsi    proc logistic data=dataset1 outest = betas covout; model SubA(event = '1') = Temp Salinity Turbidity Conductivity TC EC / selection = stepwise slentry = 0.5 slstay = 0.5 details lackfit; by Season; run; There are at least 3 issues in your problem. [1] the definition of dependent variable. For LR, it takes 0/1. You are mixing it with 5 Areas. I recommend you to use a new variable to hold 0 or 1 and use 5 areas as categorical independent variable. [2] The BY statement. The regression coefficients are determined using the number of covariate-patterns of the independent variables. It means the number of distinct patterns found in your data. The more that number better the betas. The use of BY might reduce that number. My suggestion is use SEASON as a categorical independent variable. [3] Stepwise Regression. There are two views on its need. Do not use it because SAS provides that option. Do LR first without it and if you have sufficient number of covariate patterns then try STEPWISE.   All the best. ... View more

@yashpande      I consider loading data from Agent as described below into a hash table seems promising. @Kurt_Bremser suggested other ways including Datastep Merge. Merge requires sorting of both the data sets which in your circumstance is very expensive. The length of 200 is too large to go into a hash table. If you are sure of 200 bytes for Policy_Number, you can use md5() function to bring it down to 16 bytes to save memory space for KEY-Part. Further, you can use Observation number (Record ID) of Agent as DATA-Part of Hash table. When you find a match with your LOOKUP Data set, you can recover your 36 columns of Agent by using POINT= option of Set statement. It is a challenging problem. ... View more

Here, Agent table has 8731382 records and no duplicates It has 36 variables. My B table is pain it has 167171496 and has duplicates on the key column and has total 29 columns. Both of these tables are SAS tables and have indexes on respective columns. I tried with HASH join but it still takes huge time. Do you want the output to have 36 + 167 variables? What is length of POLICY_NUMBER  ? From the use of UPCASE() I understand that POLICY_Number is not digital string.  ... View more

@YangYY    I tried again it still doesn't work. Can you please try to copy my code and use the data "combined" to try your code again and see if it work? data world_attr; set mapsgfk.world_attr; run; proc sort data=world_attr; by IDNAME; run; data world_attr; set world_attr; rename IDNAME=country; run; data combined; merge world_attr co2_emission; by country; run;   I looked at the data for 176 countries in the Excel sheet giving CO2 Emissions. I don't see world_attr data set. You have not told the location for that.  Further, Excel sheet, I do not see "American Samoa" as you have given in your first post.    I insist that you closely examine your input data and DO the homework before asking further questions. Show your log for all your steps. If it is voluminous use sample of 20 rows. I am not your programmer to do all.   Good luck for finding the solution. your work.  ... View more

@YangYY    Both the programs work fine. I suspect your input dataset. Proc Print your dataset for visualization and check for any inconsistency, particularly for the DELIMITERs.    I was curious and tried your example data set. It works fine. See the output:       ... View more

@YangYY    The revised code was not tested. Here is the solution:   data want; set have; array k _numeric_; if nmiss(of _numeric_ ) = dim(k) then delete; run; Also, the code shown by @Astounding can be adapted as:   data want; set have; if n(of _numeric_) NE 0; run; Hope this solves your difficulty. All the best. ... View more

@YangYY    May be your specification is not understood. Let us work with an example.   data have; input x :$8. y1 y2 y3; datalines; AAAAA 10 20 30 BBBBB 10 . . CCCCC . . . ; run; In this dataset, I understood that you want the first two rows only. So my code with suggestion made by @Astounding :   data want; set have; if nmiss(of _numeric_ ) = dim(k) then delete; run; produces the first two records.    Can you explain your issue using this example?     ... View more

@jeremy4    There are several other ways to solve your problem. It is known as "Table Lookup" problem. The simplest way for a beginner is what @Reeza has given.    What is the data type of 'account_id'? Is it a Number or Character. If it is Number, use of Array might be the fastest way for your problem. Please let us know. ... View more