topic Re: How to calculate the number of continuous missing values in Statistical Procedures

How to calculate the number of continuous missing values

hua — Fri, 18 Sep 2015 03:41:20 GMT

I have a dataset like following, and I want to use proc expand to interpolate the missing value if the number of continuous missing is less than 6. So I need to calculate the number of continuous missing values, and add a new variable as the number of continuous missing values to the dataset. I have 2000 stations and 30000 obs for each station, so a macro or a loop may be used to calculate.

input dataset:

station year month day var

1 2015 1 1 54

1 2015 1 2 .

1 2015 1 3 32

1 2015 1 4 48

1 2015 1 5 52

1 2015 1 6 .

1 2015 1 7 .

1 2015 1 8 .

1 2015 1 9 49

1 2015 1 10 50

2 2015 1 1 .

2 2015 1 2 53

2 2015 1 3 .

2 2015 1 4 .

2 2015 1 5 55

2 2015 1 6 .

2 2015 1 7 .

2 2015 1 8 .

2 2015 1 9 47

2 2015 1 10 58

and I want to get:

station year month day var missing

1 2015 1 1 54 0

1 2015 1 2 . 1

1 2015 1 3 32 0

1 2015 1 4 48 0

1 2015 1 5 52 0

1 2015 1 6 . 3

1 2015 1 7 . 3

1 2015 1 8 . 3

1 2015 1 9 49 0

1 2015 1 10 50 0

2 2015 1 1 . 1

2 2015 1 2 53 0

2 2015 1 3 . 2

2 2015 1 4 . 2

2 2015 1 5 55 0

2 2015 1 6 . 3

2 2015 1 7 . 3

2 2015 1 8 . 3

2 2015 1 9 47 0

2 2015 1 10 58 0

Thank you for help! 🙂

Re: How to calculate the number of continuous missing values

Ksharp — Fri, 18 Sep 2015 04:50:42 GMT

Try this one :

data have;
input inputstation  year  month  day   var;
cards;
1          2015     1        1      54
1          2015     1        2      .
1          2015     1        3      32
1          2015     1        4      48
1          2015     1        5      52
1          2015     1        6      .
1          2015     1        7      .
1          2015     1        8      .
1          2015     1        9      49
1          2015     1       10     50
2          2015     1        1      .
2          2015     1        2      53
2          2015     1        3      .
2          2015     1        4      .
2          2015     1        5      55
2          2015     1        6      .
2          2015     1        7      .
2          2015     1        8      .
2          2015     1        9      47
2          2015     1       10     58
;
run;
data want;
count=0;
do until(last.var);
 set have;
 by inputstation  var notsorted;
 if missing(var) then count+1;
end;
do until(last.var);
set have;
 by inputstation  var notsorted;
 output;
end;
run;

Xia Keshan

Re: How to calculate the number of continuous missing values

hua — Fri, 18 Sep 2015 13:39:08 GMT

Thanks very much! It works very well.

what does "var notsorted" in " by station var notsorted;" means?

Re: How to calculate the number of continuous missing values

hua — Fri, 18 Sep 2015 16:37:27 GMT

And do you know how to use if statement in proc expand process?

Re: How to calculate the number of continuous missing values

Ksharp — Sat, 19 Sep 2015 04:05:52 GMT

It take every side by side as a group. E.X.
Group
1 2015 1 4 48 1
1 2015 1 5 52 2
1 2015 1 6 . 3
1 2015 1 7 . 3
1 2015 1 8 52 4

Re: How to calculate the number of continuous missing values

Ksharp — Sat, 19 Sep 2015 04:08:10 GMT

NO. I don't think you can use IF in proc expand.
But you can firstly use proc expand and padding these missing value, then use IF in data step to process what you want.