Reshaping data from long to wide using counts

[hwangnyc]

Hi everyone,

I have a dataset that has 3 columns:

Date - Date of the event

Car - the number of the car

Client - The person who is riding in the car

If on a given date, a car has more than one client, they have two rows. I would like to make the data wide so I can get the average per car over a period of time. 

The data that I have is this:

Date	Car	Client
01/01/18	1	Ente
01/02/18	1	Patterson
01/02/18	1	Jeter
01/02/18	2	Davidson
01/03/18	1	Steve
01/03/18	1	Gonzalez
01/03/18	2	York
01/03/18	2	Camara
01/03/18	2	Camara
01/04/18	1	Waters
01/04/18	1	Ventre
01/04/18	2	Perez
01/04/18	3	Rodriguez
01/04/18	3	Jeter
01/04/18	3	Jiminez

The data that I would like is this:

Date	CAR1_Count	Car2_Count	Car3_Count	TotalCount
1/1/2018	1	0	0	1
1/2/2018	2	1	0	3
1/3/2018	2	3	0	5
1/4/2018	2	1	3	6

Any suggestions would be greatly appreciated.

Thank you!

[VDD]

data have;
infile cards missover dlm=' ';
input Date $ Car Client $;
cards;
01/01/18 1 Ente
01/02/18 1 Patterson
01/02/18 1 Jeter
01/02/18 2 Davidson
01/03/18 1 Steve
01/03/18 1 Gonzalez 
01/03/18 2 York
01/03/18 2 Camara
01/03/18 2 Camara
01/04/18 1 Waters
01/04/18 1 Ventre
01/04/18 2 Perez
01/04/18 3 Rodriguez
01/04/18 3 Jeter
01/04/18 3 Jiminez
;

proc freq data=have ;
	tables date*car / out=want;
run;
proc sql;
	create table want2 as
	select date, count(date) as TotalCount
	from have
	group by date;
quit;
proc sort data=want;
by date car;
run;
data want3;
	merge want2 want;
	by date;
run;
proc transpose data=want3(drop=percent) out=want4(drop=_name_ _label_) prefix=Count_car ;
	id car;
	by date totalcount;
run;

[Astounding]

PROC TABULATE should be able to do that easily.  Untested but definitely in the right neighborhood:

proc tabulate data=have;

class date car;

tables date, (car all)*n=' ';

run;

[Reeza]

Why do you think you need it wide to get summary statistics? Wouldn't PROC MEANS with some CLASS or formats work just as well?

[ballardw]

This uses Proc means to display the mean number of riders per car per day that the car was used.

data have;
infile cards missover dlm=' ';
input Date $ Car Client $;
cards;
01/01/18 1 Ente
01/02/18 1 Patterson
01/02/18 1 Jeter
01/02/18 2 Davidson
01/03/18 1 Steve
01/03/18 1 Gonzalez 
01/03/18 2 York
01/03/18 2 Camara
01/03/18 2 Camara
01/04/18 1 Waters
01/04/18 1 Ventre
01/04/18 2 Perez
01/04/18 3 Rodriguez
01/04/18 3 Jeter
01/04/18 3 Jiminez
;

proc summary data=have nway;
   class date car;
   output out=havesum (drop=_type_);
run;

proc means data=havesum (rename=(_freq_=count)) mean;
   class car;
   var count;
run;
   
 

If you actually need a wider period of time you may need to include the days with 0 riders depending on what you mean by average per car over a period of time.