Efficient way to keep only highest ranked observation for each group - sas

I want to keep only the row with the highest rank1 for each team. If there is a tie, I want the row with the higher rank2. And then the higher rank3.
For example,
data test;
input name $ team $ rank1 rank2 rank3 country $
datalines;
Bob A 5 6 5 US
Joe A 8 2 6 UK
Dav B 9 7 2 GER
Jim B 9 4 4 FRA
Bob C 3 4 1 FRA
Dan D 5 2 7 GER
Ike D 5 2 7 US
Jay D 5 2 8 UK
run;
I want:
Joe A 8 2 6 UK
Dav B 9 7 2 GER
Bob C 3 4 1 FRA
Jay D 5 2 8 UK
What is the most efficient way to do this? The dataset I'm working with is pretty big and is not sorted. I tried the below code but the sorts take forever to run. And the second sort sorts already sorted data. What if most teams only appear once in the dataset? Is it faster to split into duplicates and non-duplicates, sort only the duplicates and then append?
proc sort data=test;
by team descending rank1 descending rank2 descending rank3;
run;
proc sort data=test nodupkey;
by team;
run;

You can do that with PROC SUMMARY. Not sure about performance compared to what you are already doing.
proc summary data=test nway;
class team;
output out=ranked(drop=_:) idgroup(max(rank:) out(name rank: country)=);
run;

Related

Rolling Window Model for Unbalanced Dataset in SAS

I have an unbalanced panel dataset of the following form (simplified):
data have;
input ID YEAR EARN LAG_EARN;
datalines;
1 1960 450 .
1 1961 310 450
1 1962 529 310
2 1978 10 .
2 1979 15 10
2 1980 8 15
2 1981 10 8
2 1982 15 10
2 1983 8 15
2 1984 10 8
3 1972 1000 .
3 1973 1599 1000
3 1974 1599 1599
;
run;​
I now want to estimate the following model for each ID:
proc reg;
by ID;
EARN = LAG_EARN;
run;
However, I want to do this for rolling windows of some size. Say for example for windows of size 2. The window should only contain non-empty observations. For example, in the case of firm A, the window is applicable from 1961 onwards and thus only one time (since only one year follows after 1961 and the window is supposed to be of size 2).
Finally, I want to get a table with year columns and firm rows. The table should indicate the following: The regression model (with window size 2) has been performed one time for firm A. The quantity of available years, has only allowed one estimation of this model. Put differently, in 1962 the coefficient of the regression model has a value of X based on the 2 year prior window. Applying the same logic to the other two firms, one can get the following table. "X" representing the respective estimated coefficient value in certain year for firm A/B/C based on the 2-year window and "n" indicating the non-existence of such a value:
data want;
input ID 1962 1974 1980 1981 1982 1983 1984;
datalines;
1 X n n n n n n
2 n n X X X X X
3 n X n n n n n
;
run;​
I do not know how to execute this. Furthermore, I would like to create a macro that allows me to estimate different rolling window models while still creating analogous output dataframes. I would appreciate any help with it, since I have been struggling quite some time now.
Try this macro. This will only output if there are non-missing values of lags that you specify.
%macro lag(data=, out=, window=);
data _want_;
set &data.;
by ID;
LAG_EARN = lag&window.(earn);
if(first.ID) then call missing(lag_earn);
if(NOT missing(lag_earn));
run;
proc sort data=_want_;
by year id;
run;
proc transpose data=_want_
out=&out.(drop=_NAME_);
by ID notsorted;
id year;
var lag_earn;
run;
proc sort data=&out.;
by id;
run;
%mend;
%lag(data=have, out=want, window=1);

Why does the Compress function not work in SAS Enterprise Guide?

I am currently restructuring my package from SAS Base to SAS Enterprise Guide in a knowledge transfer to a client. Unfortunately, one aspect I have to sacrifice is the change from using compress to strip in my proc sql left joins, for example the following code doesn't work
data have;
input ID VarA;
datalines;
1 2
2 3
3 4
4 5
;
run;
data have1;
input ID Var1 Var2 Var3 Var4 Var5 Var6 Var7 Var8 Var9;
datalines;
1 3 4 6 7 3 6 6 7 8
2 2 2 2 2 5 6 7 2 1
3 5 6 7 8 4 5 3 4 3
4 3 4 6 7 4 6 8 3 6
;
run;
proc sql;
create table Want as
select a.*
,b.Var1
,b.Var2
,b.Var3
,b.Var4
,b.Var5
,b.Var6
,b.Var7
,b.Var8
,b.Var9
from Have as a
left join Have1 as b
on compress(a.ID) = compress(b.ID);
quit;
I can use the strip function at times but it is safer to deliver a package with compress as there is often misplaced spaces in observations. any ideas?
Edit: to save further confusion, I usually use the compress function to look up reference rates of bonds like EURIBOR 006m - this makes my generic example incorrect but the left join typically uses character variables
You need a character variable to use the compress function. Your ID variables are numeric.
Try converting to character:
on compress(put(a.ID,8.)) = compress(put(b.ID,8.));

Stacking a dataset

What's the code program in SAS to stack data?
For the purpose of example, lets say I have this dataset:
DATA test.one;
INPUT Name $ Y1996 Y1997 Y1998 Y1999;
cards;
Dan 5 10 40 20
Derek 10 12 10 10
run;
proc print data = test.one;
run;
Running this set would give me an output like this:
Name Y1996 Y1997 Y1998 Y1999
Dan 5 10 40 20
Derek 10 12 10 10
However, I would want my data to look like this:
Name Year Income
Dan 1996 5
Dan 1997 10
Dan 1998 40
Dan 1999 20
Derek 1996 10
Derek 1997 12
Derek 1998 10
Derek 1999 10
It would create a new variable income corresponding to the stacking the of the data as shown above.
Are you asking how to read the raw data directly into that form?
DATA want;
INPUT Name $ #;
do year=1996 to 1999;
input income #;
output;
end;
cards;
Dan 5 10 40 20
Derek 10 12 10 10
;
The PROC Transpose can solve this;
DATA test.one;
INPUT Name $ y1996 y1997 y1998 y1999;
cards;
Dan 5 10 40 20
Derek 10 12 10 10
run;
proc print data = test.one;
run;
proc transpose data=test.one out=long1;
by name;
run;
data test2;
set long1 (rename=(col1=Income));
RUN;
It will then transform the dataset into a stacked version.

Creating All Possible Combinations in a Table Using SAS

I have a table with four variables and i want the table a table with combination of all values. Showing a table with only 2 columns as an example.
NAME AMOUNT COUNT
RAJ 90 1
RAVI 20 4
JOHN 30 5
JOSEPH 40 3
The following output is to show the values only for raj and the output should be for all names.
NAME AMOUNT COUNT
RAJ 90 1
RAJ 90 4
RAJ 90 5
RAJ 90 3
RAJ 20 1
RAJ 20 4
RAJ 20 5
RAJ 20 3
RAJ 30 1
RAJ 30 4
RAJ 30 5
RAJ 30 3
RAJ 40 1
RAJ 40 4
RAJ 40 5
RAJ 40 3
.
.
.
.
There are a couple of useful options in SAS to do this; both create a table with all possible combinations of variables, and then you can just drop the summary data that you don't need. Given your initial dataset:
data have;
input NAME $ AMOUNT COUNT;
datalines;
RAJ 90 1
RAVI 20 4
JOHN 30 5
JOSEPH 40 3
;;;;
run;
There is PROC FREQ with SPARSE.
proc freq data=have noprint;
tables name*amount*count/sparse out=want(drop=percent);
run;
There is also PROC TABULATE.
proc tabulate data=have out=want(keep=name amount count);
class name amount count;
tables name*amount,count /printmiss;
run;
This has the advantage of not conflicting with the name for the COUNT variable.
Try
PROC SQL;
CREATE TABLE tbl_out AS
SELECT a.name AS name
,b.amount AS amount
,c.count AS count
FROM tbl_in AS a, tbl_in AS b, tbl_in AS c
;
QUIT;
This performs a double self-join and should have the desired effect.
Here's a variation on #JustinJDavies's answer, using an explicit CROSS JOIN clause:
data have;
input NAME $ AMOUNT COUNT;
datalines;
RAJ 90 1
RAVI 20 4
JOHN 30 5
JOSEPH 40 3
run;
PROC SQL;
create table combs as
select *
from have(keep=NAME)
cross join have(keep=AMOUNT)
cross join have(keep=COUNT)
order by name, amount, count;
QUIT;
Results:
NAME AMOUNT COUNT
JOHN 20 1
JOHN 20 3
JOHN 20 4
JOHN 20 5
JOHN 30 1
JOHN 30 3
JOHN 30 4
JOHN 30 5
...

Summarizing data in SAS across groups

My data set is in this format as mentioned below:
NEWID
Age
H_PERS
Income
OCCU
FAMTYPE
REGION
Metro(Yes/No)
Exp_alcohol
population sample-(This is the weighted population each new id represents) etc.
I would like to generate a summarized view like below:
average expenditure value (This should be sum of (exp_alcohol/population sample))
% of population sample across Region Metro and each demographic variable
Please help me with your ideas.
Since I can't see your data set and your description was not very clear, I'm going to guess that you have data that looks something like this and you would like add some new variables that summarizes your data...
data alcohol;
input NEWID Age H_PERS Income OCCU $ FAMTYPE $ REGION $ Metro $
Exp_alcohol population_sample;
datalines;
1234 32 4 65000 abc m CA Yes 2 4
5678 23 5 35000 xyz s WA Yes 3 6
9923 34 3 49000 def d OR No 3 9
8844 26 4 54000 gdp m CA No 1 5
;
run;
data summar;
set alcohol;
retain TotalAvg_expend metro_count total_pop;
Divide = exp_alcohol/population_sample;
TotalAvg_expend + Divide;
total_pop + population_sample;
if metro = 'Yes' then metro_count + population_sample;
percent_metro = (metro_count/total_pop)*100;
drop NEWID Age H_PERS Income OCCU FAMTYPE REGION Divide;
run;
Output:
Exp_ population_ TotalAvg_ metro_ total_ percent_
Metro alcohol sample expend count pop metro
Yes 2 4 0.50000 4 4 100.000
Yes 3 6 1.00000 10 10 100.000
No 3 9 1.33333 10 19 52.632
No 1 5 1.53333 10 24 41.667