predicting variable value

[riya_lopinti]

Hi,

I'm completly new to SAS .kindly bear with me !

so here is my question..how can i predict a value of a particular variable based on another variable value.

for instace : i have a dataset with variables risk, age and it has values as follows:

12 57

24 67

13 58

56 86

so on..

So now i need to predict the value of the risk if age increases by 10 years  or i can say

'By how much does the risk increase if age increases by ten years?'

any help is highly appreciated.Thank you 

[RW9]

I believe you really need to speak with your statistician or who requested this as there are many models and methodolgies which could be implemented for this from a simple, take base number and increment it by value * 100 + (age / 100 * 10), i.e. the percentage difference between age and age + 10 multiplied by original added on, to far more complex routines and models.  Or it could be risk doesn't increase, hence previous value remains.  Only the person who wants this can really tell you.

[SuryaKiran]

It depends on how do you want to interpolate your values. For me looking at the sample data linear interpolation would work. There are different approached in SAS to handle like this and my favorite one is using PROC EXPAND. 

Eg: I want to interpolate risk for 78(age) then I would do like this:

infile datalines dlm=" " dsd;
input risk age;
datalines;
12 57
24 67
. 78
13 58
56 86
;
run;
proc sort data=have;
by age risk;
run;
proc expand data=have out=want;
convert risk=Predicting/method=join;
id age;
run;

Also check this http://support.sas.com/kb/24/560.html

[riya_lopinti]

Hi , 

Thank you for your valuable input .However ,i'm using SAS university edition and i dont think i can use the mentioned procedure[Expand]..So is there an alternate Proc i can use to fix this issue. I just completed  SAS training and using University edition for practise purpose.So please help me on this .Thank you so much !! 

Regards

Riya.

[SuryaKiran]

Your missing values can  be filled by taking the average from previous and next values, 

data want;
retain _risk;
do i=1 by 1 until(not missing(risk_));
	set have(rename=risk=risk_);
	end;
do j=1 to i;
set have;
If not missing(risk) then _risk=risk;
if missing(risk) then risk=(_risk+risk_)/2;
Output;
end;
keep age risk;
run;

[riya_lopinti]

Hi, 

Thanks a lot for helping me .But still i'm not clear on this .So please bear with me ..

 My dataset dont have any missing values as shown below :

It has 2 variables risk, age .

12 57

24 67

13 58

56 86

28 59

51 76

Now i need to predict the value of the risk if age increases by 10 years  or simply

'By how much does the risk increase if age increases by ten years?'

Does your answer still holds good for this question.If so .please briefly explain sir.  if not please provide your suggestions .

Thank you so much !

Regards

Riya