## Create new variable based on threshold

Solved
Super Contributor
Posts: 328

# Create new variable based on threshold

[ Edited ]

I have this longitudinal data
id  PD
01  0.5
01  1.0
01  1.5
01  0.9
01  1.6
01  1.1
02  1.5
02  1.0
02  1.5
02  1.5
02  1.7
03  1.2
03  1.5
03  1.0
03  1.5
03  0.5
03  1.7
03  1.2
04  0.2
04  0.5
04  0.6
04  0.5
04  0.5
04  0.7
04  0.2

For each id, the PD value fluctuate between 0 and 1.7.

I have this three categories: i) PD>1.15 , ii) 0.9<PD<=1.15 iii) PD<=0.9
For some id, PD will fluctuate within the three groups. I want to create a third variable GroupPD and call such group "A", that is the id will
have correspondent GroupPD value "A". b) if PD fluctuate between PD>1.15 and 0.9<PD<=1.15 the GroupPD value will be "B". c) if PD fluctuate between PD>1.15 and PD<=0.9
the GroupPD value will be "C" d) if PD fluctuate between 0.9<PD<=1.15 and PD<=0.9 GroupPD value will be "D" e) If PD stayed within PD>1.15, then GroupPD will be "E"
f) If PD stayed within <0.9PD<=1.15, then GroupPD will be "F" g) If PD stayed within PD<=0.9, then GroupPD will be "G"

Please can I have code to help me do this.

Expected output

id  PD  GroupPD
01  0.5 A
01  1.0  A
01  1.5  A
01  0.9  A
01  1.6  A
01  1.1  A
02  1.5 B
02  1.0 B
02  1.5 B
02  1.5 B
02  1.7 B
03  1.2 C
03  1.5 C
03  1.0 C
03  1.5 C
03  0.5 C
03  1.7 C
03  1.2 C
04  0.2 G
04  0.5 G
04  0.6 G
04  0.5 G
04  0.5 G
04  0.7 G
04  0.2 G

Accepted Solutions
Solution
‎06-20-2017 11:51 PM
Frequent Contributor
Posts: 118

## Re: Create new variable based on threshold

``````
data have;
input id \$ 1-2 pd;
datalines;
01  0.5
01  1.0
01  1.5
01  0.9
01  1.6
01  1.1
02  1.5
02  1.0
02  1.5
02  0.5
02  1.7
03  1.2
03  1.5
03  1.0
03  1.5
03  0.5
03  1.7
03  1.2
04  0.2
04  0.5
04  0.6
04  0.5
04  0.5
04  0.7
04  0.2
;
run;

proc print data=have;
run;

* sort data set if not already sorted by id;

data want(drop=low medium high pd);
retain low medium high;
set have;
by id;
if first.id then
do;
low=0;
medium=0;
high=0;
end;
if pd le 0.9 then low+1;
else if 0.9 lt pd le 1.15 then medium+1;
else if pd gt 1.15 then high+1;
if last.id then
do;
groupPD='A';
if (low=0 and medium gt 0 and high gt 0) then groupPD='B';
else if (low gt 0 and medium=0 and high gt 0) then groupPD='C';
else if (low gt 0 and medium gt 0 and high=0) then groupPD='D';
else if (low=0 and medium=0 and high gt 0) then groupPD='E';
else if (low=0 and medium gt 0 and high=0) then groupPD='F';
else if (low gt 0 and medium=0 and  high=0) then groupPD='G';
output;
end;
run;

proc print data=want;
run;
``````

All Replies
Posts: 1,147

## Re: Create new variable based on threshold

Please show us the expected output for better response

Thanks,
Jag
Solution
‎06-20-2017 11:51 PM
Frequent Contributor
Posts: 118

## Re: Create new variable based on threshold

``````
data have;
input id \$ 1-2 pd;
datalines;
01  0.5
01  1.0
01  1.5
01  0.9
01  1.6
01  1.1
02  1.5
02  1.0
02  1.5
02  0.5
02  1.7
03  1.2
03  1.5
03  1.0
03  1.5
03  0.5
03  1.7
03  1.2
04  0.2
04  0.5
04  0.6
04  0.5
04  0.5
04  0.7
04  0.2
;
run;

proc print data=have;
run;

* sort data set if not already sorted by id;

data want(drop=low medium high pd);
retain low medium high;
set have;
by id;
if first.id then
do;
low=0;
medium=0;
high=0;
end;
if pd le 0.9 then low+1;
else if 0.9 lt pd le 1.15 then medium+1;
else if pd gt 1.15 then high+1;
if last.id then
do;
groupPD='A';
if (low=0 and medium gt 0 and high gt 0) then groupPD='B';
else if (low gt 0 and medium=0 and high gt 0) then groupPD='C';
else if (low gt 0 and medium gt 0 and high=0) then groupPD='D';
else if (low=0 and medium=0 and high gt 0) then groupPD='E';
else if (low=0 and medium gt 0 and high=0) then groupPD='F';
else if (low gt 0 and medium=0 and  high=0) then groupPD='G';
output;
end;
run;

proc print data=want;
run;
``````
Super Contributor
Posts: 328