6.6.5. statsmodels.sandbox.panel.sandwich_covariance_generic¶
covariance with (nobs,nobs) loop and general kernel
This is a general implementation that is not efficient for any special cases. kernel is currently only for one continuous variable and any number of categorical groups.
No spatial example, continuous is interpreted as time
Created on Wed Nov 30 08:20:44 2011
Author: Josef Perktold License: BSD-3
6.6.5.1. Functions¶
S_all_hac(x, d[, nlags]) |
HAC independent of categorical group membership |
S_cluster(x, d[, groupidx]) |
|
S_white(x, d) |
simple white heteroscedasticity robust covariance |
S_within_hac(x, d[, nlags, groupidx]) |
HAC for observations within a categorical group |
aggregate_cov(x, d[, r, weights]) |
sum of outer procuct over groups and time selected by r |
kernel(d1, d2[, r, weights]) |
general product kernel |
weights_bartlett(nlags) |