About Franck1048

Franck1048 · ‎04-18-2024

We are currently in the process of evaluating the migration from SAS Base 9.4 to SAS Viya with cloud computing. If I understand correctly, the only coding interface available in SAS Viya is SAS Studio. Now, I tried SAS Studio, and unless I'm missing something, the configuration options are quite limited. I can't reproduce a window arrangement to see the explorer, code, log and result viewer at the same time. Running code and switching between tabs to see the log and results seems inefficient. Furthermore, SAS Base might not be the prettiest interface but it's more responsive. I attached a screenshot of an example of window arrangement we use for coding. Is there any way to setup the same workflow with SAS Viya? Thank you

Franck1048 · ‎01-15-2019

Ok I understand. Then yes across is definitely part of the solution. I've still yet to figure out if it's possible to output a single table using two different vars with the same format for grouping.

Franck1048 · ‎01-14-2019

I'm not sure this is what I'm trying to do. It looks like it, but after reading it's not the same. Across splits a single variable into columns. For my needs, it's my grouping variables that are different. tx_couv_cour_arrond and tx_couv_proj_arrond are two different variables that I make groups with using a common format. So one observation could be in group 01 for tx_couv_cour_arrond and in group 04 for tx_couv_proj_arrond.

Franck1048 · ‎01-09-2019

I'm producing these two tables with proc report: Using this code : proc report data=d_result.f_medo_tbc; title "Soldes courants - &DemogAnneeCourante"; column tx_couv_cour_arrond Solde1_1GRegie_Cour,(n pctn) Solde1_1GRegie_Cour=cour_sum ; define tx_couv_cour_arrond / group format=taux_de_couverture.; define pctn / FORMAT=NLPCTN8.1; define cour_sum / analysis sum f=6.0; run; proc report data=d_result.f_medo_tbc; title "Soldes courants - &DemogAnneeProjetee"; column tx_couv_proj_arrond Solde1_2GRegie_2021,(n pctn) Solde1_2GRegie_2021=proj_sum ; define tx_couv_proj_arrond / group format=taux_de_couverture.; define pctn / FORMAT=NLPCTN8.1; define proj_sum / analysis sum f=6.0; run; I want to get this result (doctored image) : As you can see, the format used for grouping is the same (taux_de_couverture.), but the variables differ. I tried combining the two procs into a single one... proc report data=d_result.f_medo_tbc; title "Soldes courants - &DemogAnneeCourante"; column tx_couv_cour_arrond Solde1_1GRegie_Cour,(n pctn) Solde1_1GRegie_Cour=cour_sum tx_couv_proj_arrond Solde1_2GRegie_2021,(n pctn) Solde1_2GRegie_2021=proj_sum ; define tx_couv_cour_arrond / group format=taux_de_couverture.; define pctn / FORMAT=NLPCTN8.1; define cour_sum / analysis sum f=6.0; define tx_couv_proj_arrond / group format=taux_de_couverture.; define pctn / FORMAT=NLPCTN8.1; define proj_sum / analysis sum f=6.0; run; ..but as expected the grouping simply gets divided into further groups : I tried a solution using proc tabulate but I'm also stuck with two procs. Any ideas? Thank you!

Franck1048 · ‎10-05-2018

Ok it worked with this : proc tabulate data=region; class flag year region; var Correction; table year * flag, region * (Correction*Sum*f=8.2 pctn<flag all>) all * (Correction*Sum*f=8.2 pctn<flag all>) ; run; That exactly the result I want. Thank you Reeza!

Franck1048 · ‎10-05-2018

It works with this method, but we don't need the weight statement here, otherwise it applies the weight "twice". So the solution would be : proc tabulate data=region; class flag year region; var Correction; table year * flag, region * (Correction*Sum*f=8.2 pctn<flag all>) all; run; The only problem with this table is the Total column, which displays the N. Any idea how to put the total of the sums instead? Thank you!!

Franck1048 · ‎10-05-2018

I'm trying to go from a simple two-way frequency table to a proc tabulate. However my weight value uses decimals (1.0677549165). This works fine with proc freq : data region; infile datalines ; input flag:$1. year:$4. region:$8.; Correction = 79867/74799; datalines; Y 2018 Region01 Y 2017 Region02 N 2017 Region01 Y 2017 Region01 N 2018 Region02 Y 2017 Region01 Y 2017 Region01 Y 2017 Region01 N 2018 Region01 Y 2018 Region01 Y 2018 Region02 N 2018 Region02 Y 2017 Region01 N 2017 Region02 Y 2017 Region01 N 2018 Region01 Y 2018 Region02 N 2017 Region01 Y 2017 Region01 Y 2018 Region01 N 2017 Region01 N 2018 Region01 N 2017 Region01 N 2018 Region02 N 2017 Region02 N 2018 Region02 Y 2018 Region02 Y 2017 Region01 Y 2018 Region01 Y 2018 Region01 ;;;; * Flag by region, flag by year; proc freq data=region; table flag * region / nocum norow nopercent; table flag * year / nocum norow nopercent; weight Correction; run; This way I get two tables, one per region and one per year with the weight properly taken into account : Now I would like to have a single table crossing both region and year. I would do this with proc tabulate : proc tabulate data=region; class flag year region; table year * flag, region * (N pctn<flag all>) all; freq Correction; run; The result is not weighted, because the FREQ option removes the decimals of Correction internally, in this case using the value of 1 : Any idea how I should proceed? Thank you

Franck1048 · ‎09-07-2018

Yeah I can see it's a lot cleaner. The _N columns clutter the second option way too much. Thank you for your help

Franck1048 · ‎09-07-2018

I can indeed replace the missing with zeros, as I have control over the input data. Using proc means I get this result : Obs. Variable N Sum Mean 1 error1 10 5 50,00% 2 error2 10 7 70,00% It gives me the results I want, except in a row format, instead of columns. I tried different methods to get the output in columns but I can't wrap my head around it.

Franck1048 · ‎09-07-2018

My data has columns with values that are either missing or 1 : data test; input id error1 error2; cards; 23 1 . 2 . 1 345 1 1 3 1 . 45 1 . 67 . 1 789 . 1 89 . 1 34 . 1 21 1 1 ; I want to output a report that counts the number of error1 occurences and its percentage of the total observations. I also need the report to output the same information for error2, error3, error4.. and so on if they exist .. basically for error* regardless of the number of error columns. Example of result I'm looking for : All error1 error2 error3.. Count Count Percent Count Percent Count Percent 10 5 50% 7 70% Thank you

Online Status	Offline
Date Last Visited	‎04-19-2024 12:58 PM

Windowed workflow for coding - SAS Base vs SAS Studio

Re: Combine two proc report that use the same categorization format

Re: Combine two proc report that use the same categorization format

Combine two proc report that use the same categorization format

Re: Proc TABULATE : FREQ option ignores decimals

Re: Proc TABULATE : FREQ option ignores decimals

Proc TABULATE : FREQ option ignores decimals

Re: Calculate sums and percentages for a variable number of columns (r...

Re: Calculate sums and percentages for a variable number of columns (r...

Calculate sums and percentages for a variable number of columns (range...

Re: Windowed workflow for coding - SAS Base vs SAS Studio

Re: Windowed workflow for coding - SAS Base vs SAS Studio

Re: Combine two proc report that use the same categorization format

Re: Proc TABULATE : FREQ option ignores decimals

Re: Calculate sums and percentages for a variable number of columns (r...

Windowed workflow for coding - SAS Base vs SAS Studio

Re: Combine two proc report that use the same categorization format

Re: Combine two proc report that use the same categorization format

Combine two proc report that use the same categorization format

Re: Proc TABULATE : FREQ option ignores decimals

Re: Proc TABULATE : FREQ option ignores decimals

Proc TABULATE : FREQ option ignores decimals

Re: Calculate sums and percentages for a variable number of columns (r...

Re: Calculate sums and percentages for a variable number of columns (r...

Calculate sums and percentages for a variable number of columns (range...