Identify first and last

[lizzy28]

Hi all,

I have to admit my do-loop skill is too weak. I need to sort out the first and last months when shipping was made for each year within a year. As shown below, the columns of startmon and endmon are my objective variables I want.

OrderID	mons	mon1	mon2	mon3	mon4	mon5	mon6	mon7	mon8	mon9	mon10	mon11	mon12	year	startmon	endmon
1	4	1	1	1	1	0	0	0	0	0	0	0	0	2012	1	4
2	11	0	1	1	1	1	1	1	1	1	1	1	1	2012	2	12
3	12	1	1	1	1	1	1	1	1	1	1	1	1	2012	1	12
4	1	1	0	0	0	0	0	0	0	0	0	0	0	2012	1	1
5	6	0	0	0	1	1	1	1	1	1	0	0	0	2012	4	9

Also, for the orders across years as below, I need to abstract a separate file with startmon, endmon, startyr and endyr.

Original data:

OrderID	mons	mon1	mon2	mon3	mon4	mon5	mon6	mon7	mon8	mon9	mon10	mon11	mon12	year
1	12	12	1	1	1	1	1	1	1	1	1	1	1	2011
3	3	0	0	0	0	0	0	0	0	0	1	1	1	2011
4	11	0	1	1	1	1	1	1	1	1	1	1	1	2011
1	4	1	1	1	1	0	0	0	0	0	0	0	0	2012
2	11	0	1	1	1	1	1	1	1	1	1	1	1	2012
3	12	1	1	1	1	1	1	1	1	1	1	1	1	2012
4	1	1	0	0	0	0	0	0	0	0	0	0	0	2012
5	6	0	0	0	1	1	1	1	1	1	0	0	0	2012
1	12	1	1	1	1	1	1	1	1	1	1	1	1	2013
2	10	1	1	1	1	1	1	1	1	1	1	0	0	2013
3	2	1	1	0	0	0	0	0	0	0	0	0	0	2013
5	12	1	1	1	1	1	1	1	1	1	1	1	1	2013

Objective data:

OrderID	startmon	endmon	staryr	endyr
1	1	4	2011	2012
2	2	10	2012	2013
3	10	2	2011	2013
4	2	1	2011	2012
5	4	12	2012	2013

Your ideas are greatly appreciated!

Lizi

[Reeza]

Here's the first part, first/last month. I'm not sure about the logic of first/last year? You just want the first/last year per order id? Which start/end month should align with the years?

data have;

input OrderID mons mon1 mon2 mon3 mon4 mon5 mon6 mon7 mon8 mon9 mon10 mon11 mon12 year startmon endmon;

cards;

1 4 1 1 1 1 0 0 0 0 0 0 0 0 2012 1 4

2 11 0 1 1 1 1 1 1 1 1 1 1 1 2012 2 12

3 12 1 1 1 1 1 1 1 1 1 1 1 1 2012 1 12

4 1 1 0 0 0 0 0 0 0 0 0 0 0 2012 1 1

5 6 0 0 0 1 1 1 1 1 1 0 0 0 2012 4 9

;

run;

data want;

set have;

length string $12.;

array mnth(12) mon1-mon12;

string=cats(of mon1-mon12);

first_month=find(string, '1');

last_month=12-find(reverse(string), '1')+1;

run;

[lizzy28]

Thanks a lot, Reeza.

You addressed my first question.

For the second part, I need to observe the months across several years and find the first and last months over the years as well as the first and last years of shipping per order. I think I can use your idea once I transform the data from long to wide. Thanks again!

[Ksharp · Accepted solution]

I want know if there are some non-consecutive range ,like :

data have;

input OrderID mons mon1 mon2 mon3 mon4 mon5 mon6 mon7 mon8 mon9 mon10 mon11 mon12 year startmon endmon;

cards;

1 4 1 1 1 1 0 0 0 0 1 1 1 1 2012 1 4

2 11 0 1 1 1 1 1 1 1 1 1 1 1 2012 2 12

3 12 1 1 1 1 1 1 1 1 1 1 1 1 2012 1 12

4 1 1 0 0 0 0 0 0 0 0 0 0 0 2012 1 1

5 6 0 0 0 1 1 1 1 1 1 0 0 0 2012 4 9

;

For the first one:

data have;
input OrderID mons mon1 mon2 mon3 mon4 mon5 mon6 mon7 mon8 mon9 mon10 mon11 mon12 year startmon endmon;
cards;
1 4 1 1 1 1 0 0 0 0 0 0 0 0 2012 1 4
2 11 0 1 1 1 1 1 1 1 1 1 1 1 2012 2 12
3 12 1 1 1 1 1 1 1 1 1 1 1 1 2012 1 12
4 1 1 0 0 0 0 0 0 0 0 0 0 0 2012 1 1
5 6 0 0 0 1 1 1 1 1 1 0 0 0 2012 4 9
;
run;
data temp;
 set have;
 array m{*} mon1-mon12;
 do i=1 to dim(m);
  value=m{i};output;
 end;
 keep  OrderID i value;
run;
data want;
 set temp;
 by OrderID value notsorted;
 retain start_mon ;
 if first.value then start_mon=i;
 if last.value and value=1 then do;end_mon=i;output;end;
 drop i value;
run;

For the second one:

data have;
input OrderID mons mon1 mon2 mon3 mon4 mon5 mon6 mon7 mon8 mon9 mon10 mon11 mon12 year ;
cards;
1     12     1     1     1     1     1     1     1     1     1     1     1     1     2011
3     3     0     0     0     0     0     0     0     0     0     1     1     1     2011
4     11     0     1     1     1     1     1     1     1     1     1     1     1     2011
1     4     1     1     1     1     0     0     0     0     0     0     0     0     2012
2     11     0     1     1     1     1     1     1     1     1     1     1     1     2012
3     12     1     1     1     1     1     1     1     1     1     1     1     1     2012
4     1     1     0     0     0     0     0     0     0     0     0     0     0     2012
5     6     0     0     0     1     1     1     1     1     1     0     0     0     2012
1     12     1     1     1     1     1     1     1     1     1     1     1     1     2013
2     10     1     1     1     1     1     1     1     1     1     1     0     0     2013
3     2     1     1     0     0     0     0     0     0     0     0     0     0     2013
5     12     1     1     1     1     1     1     1     1     1     1     1     1     2013
;
run;
data temp;
 set have;
 array m{*} mon1-mon12;
 do i=1 to dim(m);
  value=m{i};date=mdy(i,1,year);output;
 end;
 format date date9.;
 keep  OrderID date  value;
run;
proc sort data=temp; by OrderID date;run;
data want;
 set temp;
 by OrderID   value notsorted;
 retain start_mon start_year;
 if first.value then do;start_mon=month(date); start_year=year(date); end;
 if last.value and value=1 then do;end_mon=month(date); end_year=year(date);output;end;
 drop value;
run;

Xia Keshan

[lizzy28]

Thanks so much, Keshan! You have what I've been looking for.

Lizi

[lizzy28]

Keshan, I didn't pay attention to your comments. But yes, there can be some nonconsecutive ranges. If I only care about the start month and the end month only despite non-consecutive ranges, how would the program change? Thanks again!

[Ksharp]

You mean consecutive range's start month and end month?  If it was, keep my code intact .

[lizzy28]

I mean there can be breaks during a year. Just like OrderID#6 as below:

OrderID	mons	mon1	mon2	mon3	mon4	mon5	mon6	mon7	mon8	mon9	mon10	mon11	mon12	year
1	12	12	1	1	1	1	1	1	1	1	1	1	1	2011
3	3	0	0	0	0	0	0	0	0	0	1	1	1	2011
4	11	0	1	1	1	1	1	1	1	1	1	1	1	2011
6	5	1	1	0	0	0	0	0	0	0	1	1	1	2011
1	4	1	1	1	1	0	0	0	0	0	0	0	0	2012
2	11	0	1	1	1	1	1	1	1	1	1	1	1	2012
3	12	1	1	1	1	1	1	1	1	1	1	1	1	2012
4	1	1	0	0	0	0	0	0	0	0	0	0	0	2012
5	6	0	0	0	1	1	1	1	1	1	0	0	0	2012
6	11	1	1	1	1	1	1	1	1	1	1	1	0	2012
1	12	1	1	1	1	1	1	1	1	1	1	1	1	2013
2	10	1	1	1	1	1	1	1	1	1	1	0	0	2013
3	2	1	1	0	0	0	0	0	0	0	0	0	0	2013
5	12	1	1	1	1	1	1	1	1	1	1	1	1	2013
6	8	0	1	1	1	1	1	1	0	1	1	0	0	2013

OrderID #6 has 4 breaks, but I only need the first start, month mon1 (January) of 2011, and the last end month, month mon10 (October) of 2013

OrderID	mons	mon1	mon2	mon3	mon4	mon5	mon6	mon7	mon8	mon9	mon10	mon11	mon12	year
6	5	1	1	0	0	0	0	0	0	0	1	1	1	2011
6	11	1	1	1	1	1	1	1	1	1	1	1	0	2012
6	8	0	1	1	1	1	1	1	0	1	1	0	0	2013

So, the target data is

OrderID	startmon	startyr	endmon	endyr
1	1	2011	4	2012
2	2	2012	10	2013
3	10	2011	2	2013
4	2	2011	1	2012
5	4	2012	12	2013
6	1	2011	10	2013

Thanks for all the help, Keshan!

[Ksharp]

OK. Here is .

data have;
input OrderID mons mon1 mon2 mon3 mon4 mon5 mon6 mon7 mon8 mon9 mon10 mon11 mon12 year ;
cards;
1     12     1     1     1     1     1     1     1     1     1     1     1     1     2011
3     3     0     0     0     0     0     0     0     0     0     1     1     1     2011
4     11     0     1     1     1     1     1     1     1     1     1     1     1     2011
6     5     1     1     0     0     0     0     0     0     0     1     1     1     2011
1     4     1     1     1     1     0     0     0     0     0     0     0     0     2012
2     11     0     1     1     1     1     1     1     1     1     1     1     1     2012
3     12     1     1     1     1     1     1     1     1     1     1     1     1     2012
4     1     1     0     0     0     0     0     0     0     0     0     0     0     2012
5     6     0     0     0     1     1     1     1     1     1     0     0     0     2012
6     11     1     1     1     1     1     1     1     1     1     1     1     0     2012
1     12     1     1     1     1     1     1     1     1     1     1     1     1     2013
2     10     1     1     1     1     1     1     1     1     1     1     0     0     2013
3     2     1     1     0     0     0     0     0     0     0     0     0     0     2013
5     12     1     1     1     1     1     1     1     1     1     1     1     1     2013
6     8     0     1     1     1     1     1     1     0     1     1     0     0     2013
;
run;
data temp(where=(value=1));
 set have;
 array m{*} mon1-mon12;
 do i=1 to dim(m);
  value=m{i};date=mdy(i,1,year);output;
 end;
 format date date9.;
 keep  OrderID date  value;
run;
proc sort data=temp; by OrderID date;run;
data want;
 set temp;
 by OrderID;
 retain start_mon start_year;
 if first.OrderID then do;start_mon=month(date); start_year=year(date); end;
 if last.OrderID then do;end_mon=month(date); end_year=year(date);output;end;
 drop value;
run;

Xia Keshan

[lizzy28]

Thanks a lot, Keshan. You helped me tons.