Dear all
I have time-series data. I want to run rolling regressions and obtain the decomposed R2 (also called Shapley values)
I can easily obtain the slope coefficients as follows.
However, to get the decomposed R2, the commands that can be used would be rego, shapley, shapley2, among others.
The problem is that I do not know how rego, for example, saves the decomposed R2. I did the following to investigate further
Then I thought to try something like:
However, I get an error message. I hope someone can help me get two variables each representing the decomposed R2 for each regressor.
I have time-series data. I want to run rolling regressions and obtain the decomposed R2 (also called Shapley values)
I can easily obtain the slope coefficients as follows.
Code:
rolling _b , window (40) saving (betas, replace) keep(fqdate):reg rgdp x1 x2
The problem is that I do not know how rego, for example, saves the decomposed R2. I did the following to investigate further
Code:
. rego gdp x1 x2, noperc
------------------------------------------------------------------------------
Gr Regressor | Coef. Std.Err. P>|t| Std.Coef. Shapley R2
--------------------+---------------------------------------------------------
1 x1 | -.1608338 * .0853014 0.061 -0.1375 0.0191
2 x2 | .056379 * .0308584 0.069 0.1333 0.0179
- Intercept | .0256372 .0061803 0.000
--------------------+---------------------------------------------------------
Observations | 184
Overall R2 | 0.03703
Root MSE | .0799205
F-stat. Model | 3.480318 ** 0.033
Log Likelihood | 205.3446
------------------------------------------------------------------------------
. ereturn list
scalars:
e(N) = 184
e(df_m) = 2
e(df_r) = 181
e(F) = 3.480318446798436
e(r2) = .0370324181096345
e(rmse) = .0799205236317307
e(mss) = .0444596071032519
e(rss) = 1.156099507660175
e(r2_a) = .0263918923428902
e(ll) = 205.3446297734195
e(ll_0) = 201.8729608831572
e(rank) = 3
e(noperc) = 1
macros:
e(cmdline) : "rego gdp x1 x2, noperc"
e(regressors) : " x1 x2"
e(cmd) : "rego"
e(title) : "Linear regression"
e(marginsok) : "XB default"
e(vce) : "ols"
e(depvar) : "gdp"
e(properties) : "b V"
e(predict) : "regres_p"
e(model) : "ols"
e(estat_cmd) : "regress_estat"
matrices:
e(b) : 1 x 3
e(V) : 3 x 3
e(shapley_perc) : 1 x 2
e(shapley) : 1 x 2
e(group_details) : 2 x 1
e(vars_per_group) : 2 x 1
e(stdb) : 1 x 3
functions:
e(sample)
Then I thought to try something like:
Code:
. rolling _b decomposed=e(shapley_perc), window (40) saving (betas, replace) keep(fqdate):rego gdp x1 x2, noperc (running rego on estimation sample) type mismatch error in expression: e(shapley_perc) r(109);
Comment