import numpy as np
import pyfixest as pf
fit = pf.feols("Y ~ X1 | f1", pf.get_data(), vcov={"CRV1": "f1"})
cov = fit.variance_covariance
cov.spec.vcov_type_detail, cov.G, cov.df_t, cov.ssc('CRV1', (30,), 29, array([1.03552244]))
Variance-covariance estimate of the coefficients and its building blocks.
| Name | Type | Description | Default |
|---|---|---|---|
| vcov | NDArray[np.float64] | Small-sample-adjusted covariance matrix, shape (n_coefficients, n_coefficients). | required |
| meat | NDArray[np.float64] or None | Adjusted meat of the sandwich, shape (n_coefficients, n_coefficients), so that vcov == bread @ meat @ bread with the bread of fit.sandwich. For multiway clustering the per-dimension meats enter with their signs and adjustment factors. None where no sandwich exists: "iid", "CRV3", and quantile regression. |
required |
| ssc | NDArray[np.float64] | Small-sample adjustment factors. Length one, or one entry per cluster combination for CRV inference: 2**n_clusters - 1 entries. |
required |
| df_k | int | Number of parameters counted by the k_adj adjustment. |
required |
| df_t | int or float | Degrees of freedom of the t reference distribution. | required |
| spec | VcovSpec | The requested estimator: family, detail, cluster variables, and HAC arguments. | required |
| G | tuple[int, …] | Cluster counts per dimension after the G_df rule; empty unless clustered. |
required |
('CRV1', (30,), 29, array([1.03552244]))