Generate variables from Least angel regression (Lars) coefficients

[robert xu]

Dear Community,

I want to run Least Angle Regression (Lars) each month and get the coefficients on each variable for the minimum cp each month. I can run a loop of lars, but I don't know how to generate variables from the output of the LARS (the coefficients of variables for the minimum cp model). Can anyone help me with this issue?

See the following example:

. sysuse auto,replace
(1978 Automobile Data)

. lars price weight length mpg turn rep78 headroom trunk displacement gear_ratio foreign
NOTE: Deleting all matrices
ade[11,10]
mu[1,1]
meanx[1,10]
R2[1,11]
RSS[1,11]
r2[1,1]
rss[1,1]
cp[1,11]
normx[1,10]
beta[11,10]
sbeta[11,10]
error[1,1]

sbeta[11,10]
c1 c2 c3 c4 c5 c6 c7 c8 c9 c10
r1 0 0 0 0 0 0 0 0 0 0
r2 0 0 0 0 0 0 0 29.58997 0 0
r3 4206.5105 0 0 0 0 0 0 4236.1004 0 0
r4 8354.7246 0 0 0 0 0 0 7563.3658 0 7321.7572
r5 9083.8849 0 0 0 0 -646.21787 0 8153.5461 0 8294.1935
r6 11314.929 0 0 0 1007.6609 -2770.2882 0 10176.517 0 10796.469
r7 12723.966 0 0 -1356.9973 1030.8162 -3147.7325 0 10571.041 0 11113.29
r8 14013.567 0 0 -2568.2379 1028.7096 -3478.7636 0 10649.105 -409.21915 11560.397
r9 19421.302 -4246.1604 0 -3103.4225 1210.2002 -3494.6232 0 10048.91 -797.18484 12114.523
r10 20594.449 -5323.0166 0 -3209.572 1219.9377 -3603.9899 288.90929 9897.2915 -874.63332 12224.024
r11 29559.447 -14348.173 -1054.8395 -4185.059 1508.5046 -4471.0833 2560.6707 8864.168 -1215.7602 12746.232

Algorithm is lars

Cp, R-squared and Actions along the sequence of models

+-----------------------------------------------+
Step Cp R-square Action
------+-------------+----------+---------------
1 77.5851 0.0000
2 79.3901 0.0013 +displacement
3 43.1477 0.2658 +weight
4 17.3644 0.4580 +foreign
5 15.9537 0.4816 +headroom
6 9.9905 0.5367 +rep78
7 10.1221 0.5496 +turn
8 10.6949 0.5595 +gear_ratio
9 9.7780 * 0.5796 +length
10 11.2126 0.5835 +trunk
11 11.0000 0.5989 +mpg
+-----------------------------------------------+
* indicates the smallest value for Cp

The coefficient values for the minimum Cp

+-----------------------------+
Variable Coefficient
--------------+--------------
weight 2.9705
length -22.6365
turn -84.7424
rep78 148.2509
headroom -496.7041
displacement 13.0825
gear_ratio -208.9403
foreign 3169.5719
+-----------------------------+

.
So basically, I want to create variables from the above coefficients. If the regression is OLS, I can use "gen cof_weight=_b[weight]". But this does not work for the LARS.

Thank you in advance,
Bob

[William Lisowski]

Welcome to Statalist, Bob.

For the benefit of other members, lars is a user-written Stata command available from SSC. The output of your demonstration, when presented using CODE delimiters, is more readable.

. lars price weight length mpg turn rep78 headroom trunk displacement gear_ratio foreign
NOTE: Deleting all matrices
          ade[11,10]
           mu[1,1]
        meanx[1,10]
           R2[1,11]
          RSS[1,11]
           r2[1,1]
          rss[1,1]
           cp[1,11]
        normx[1,10]
         beta[11,10]
        sbeta[11,10]
        error[1,1]

sbeta[11,10]
             c1          c2          c3          c4          c5          c6          c7
 r1           0           0           0           0           0           0           0
 r2           0           0           0           0           0           0           0
 r3   4206.5105           0           0           0           0           0           0
 r4   8354.7246           0           0           0           0           0           0
 r5   9083.8849           0           0           0           0  -646.21787           0
 r6   11314.929           0           0           0   1007.6609  -2770.2882           0
 r7   12723.966           0           0  -1356.9973   1030.8162  -3147.7325           0
 r8   14013.567           0           0  -2568.2379   1028.7096  -3478.7636           0
 r9   19421.302  -4246.1604           0  -3103.4225   1210.2002  -3494.6232           0
r10   20594.449  -5323.0166           0   -3209.572   1219.9377  -3603.9899   288.90929
r11   29559.447  -14348.173  -1054.8395   -4185.059   1508.5046  -4471.0833   2560.6707

             c8          c9         c10
 r1           0           0           0
 r2    29.58997           0           0
 r3   4236.1004           0           0
 r4   7563.3658           0   7321.7572
 r5   8153.5461           0   8294.1935
 r6   10176.517           0   10796.469
 r7   10571.041           0    11113.29
 r8   10649.105  -409.21915   11560.397
 r9    10048.91  -797.18484   12114.523
r10   9897.2915  -874.63332   12224.024
r11    8864.168  -1215.7602   12746.232

Algorithm is lars

Cp, R-squared and Actions along the sequence of models

+-----------------------------------------------+
| Step |      Cp     | R-square |  Action       |
|------+-------------+----------+---------------|
|    1 |    77.5851  |  0.0000  |               | 
|    2 |    79.3901  |  0.0013  | +displacement | 
|    3 |    43.1477  |  0.2658  | +weight       | 
|    4 |    17.3644  |  0.4580  | +foreign      | 
|    5 |    15.9537  |  0.4816  | +headroom     | 
|    6 |     9.9905  |  0.5367  | +rep78        | 
|    7 |    10.1221  |  0.5496  | +turn         | 
|    8 |    10.6949  |  0.5595  | +gear_ratio   | 
|    9 |     9.7780 *|  0.5796  | +length       | 
|   10 |    11.2126  |  0.5835  | +trunk        | 
|   11 |    11.0000  |  0.5989  | +mpg          | 
+-----------------------------------------------+
* indicates the smallest value for Cp

The coefficient values for the minimum Cp

+-----------------------------+
| Variable     |  Coefficient |
|--------------+--------------|
| weight       |       2.9705 |
| length       |     -22.6365 |
| turn         |     -84.7424 |
| rep78        |     148.2509 |
| headroom     |    -496.7041 |
| displacement |      13.0825 |
| gear_ratio   |    -208.9403 |
| foreign      |    3169.5719 |
+-----------------------------+

. return list

matrices:
                r(RSS) :  1 x 11
                 r(R2) :  1 x 11
           r(newbetas) :  1 x 10
                 r(cp) :  1 x 11
              r(sbeta) :  11 x 10
              r(normx) :  1 x 10
               r(beta) :  11 x 10

You'll note that I added the return list command, which shows the matrices returned by lars and documented in the output of help lars, You'll find what you need among them, I believe.

With that said, please review the Statalist FAQ linked to from the top of the page, as well as from the Advice on Posting link on the page you used to create your post. Note especially sections 9-12 on how to best pose your question.

The more you help others understand your problem, the more likely others are to be able to help you solve your problem.