topic How to write a SAS for saving regression residuals ? in SAS Programming

How to write a SAS for saving regression residuals ?

srikanthyadav44 — Sat, 25 Dec 2021 15:03:26 GMT

dear all,

i have the data of 163 companies stocks returns which are regressed on market return for every year separately from 2010 to 2021. i am using the below mentioned SAS code for running the regression. I need help on saving the regression residuals in a separate file

the my input data format is as follows

Date	year	rm	C1	C2	C3	……….	C163
5/4/2010	2010	1.446447751	0.034732	-1.30527	1.414732		0.114732
6/4/2010	2010	-0.059983621	0.204732	0.124732	-0.32527		-1.17527
7/4/2010	2010	0.145802794	2.394732	0.174732	1.314732		-0.02527
8/4/2010	2010	-1.330004037	-1.26527	-2.06527	-0.22527		-0.13527
9/4/2010	2010	1.059164775	-0.48527	0.384732	-0.28527		0.164732
12/4/2010	2010	-0.428505517	-0.74641	-1.14641	-0.91641		0.613589
13/4/2010	2010	-0.330592313	-0.90641	3.443589	0.973589		1.223589
15/4/2010	2010	-0.947853211	-0.47641	0.373589	-0.59641		-1.14641
16/4/2010	2010	-0.225215255	-0.53641	-1.36641	-0.73641		0.093589
19/4/2010	2010	-1.142655323	-1.46617	-2.06617	1.773835		-0.57617
20/4/2010	2010	0.490844146	0.133835	0.743835	6.183835		1.563835
21/4/2010	2010	0.266412308	1.143835	0.083835	3.933835		2.383835
22/4/2010	2010	0.448918494	-1.31617	0.323835	1.743835		1.013835
23/4/2010	2010	0.641143665	-1.81617	0.593835	1.713835		0.183835
26/4/2010	2010	0.32919628	0.533835	0.693835	1.893835		-2.44617
27/4/2010	2010	-0.281432562	-0.52617	0.883835	1.043835		-0.87617
28/4/2010	2010	-1.781733268	0.543835	-1.99617	-2.63617		-1.67617
29/4/2010	2010	0.723121179	-1.19617	-0.86617	-0.35617		-0.38617
30/4/2010	2010	0.436734309	0.973835	-1.02617	0.103835		-0.24617
3/5/2010	2010	-1.068325706	-0.35601	1.94399	0.35399		-1.40601
4/5/2010	2010	-1.447877506	-2.72601	0.48399	-0.33601		-2.42601
5/5/2010	2010	-0.475449956	0.73399	-0.80601	0.25399		-2.13601
6/5/2010	2010	-0.682630399	-1.12601	0.29399	0.02399		-1.91601
7/5/2010	2010	-1.456350078	-1.33601	-0.30601	-0.65601		-4.86601
10/5/2010	2010	3.422403185	2.813835	0.823835	3.373835		6.043835
11/5/2010	2010	-1.128498123	-0.40617	-0.05617	0.223835		1.703835
12/5/2010	2010	0.382171782	-0.83617	0.253835	-0.16617		2.803835
13/5/2010	2010	0.41438802	1.443835	2.953835	0.343835		2.533835
14/5/2010	2010	-1.678911592	0.213835	-2.05617	2.253835		-3.73617
17/5/2010	2010	-0.678170681	-3.32632	-1.35632	-2.28632		-1.86632

In the above file, rm is market return and C1 to C163 are the returns of 163 stocks.

i need the output in the following format

Date	year	C1_residual	C2_residual	C3_residual	……….	C163_residual

i am using the following code for running the regression

proc reg data=have;
model C1 - C163= rm ;
BY year;
run;
quit;

please help me writing the code to get output in the above mentioned format

thanks in advance

Re: How to write a SAS for saving regression residuals ?

PaigeMiller — Sat, 25 Dec 2021 17:16:38 GMT

Use the OUTPUT command in PROC REG. An example is given here:

https://documentation.sas.com/doc/en/pgmmvacdc/9.4/statug/statug_reg_syntax10.htm

Re: How to write a SAS for saving regression residuals ?

srikanthyadav44 — Sun, 26 Dec 2021 04:23:51 GMT

dear
in your referred source, the following code is given
proc reg data=a;
model y z=x1 x2;
output out=b
p=yhat zhat
r=yresid zresid;
run;
in that example, only two dependent variables(y and z) were there so, it was mentioned directly.
but, for my dataset, i have 163 dependent variables. i cannot mention the residual names for all the 163 variables.
is there any alternative ????
thanks in advance

Re: How to write a SAS for saving regression residuals ?

PaigeMiller — Sun, 26 Dec 2021 11:26:51 GMT

You are allowed to use any valid SAS variable list.

r=resid_c1-resid_c163

Re: How to write a SAS for saving regression residuals ?

Sajid01 — Mon, 27 Dec 2021 12:36:18 GMT

Hello @srikanthyadav44

Considering the four c values in your data, the code shoutd be something like as shown below.
Please note the output file will be in the work library.
Modify as needed.

proc reg data=your_source_data;
model c1 c2 c3 c4 =rm ;
by year;
output out=residual_output 
p=c1hat c2hat c3hat c4hat  
r=c1resid c2resid c3resid c4resid   ;
run;
quit;