lyman.glm.iterative_ols_fit¶
-
lyman.glm.iterative_ols_fit(Y, X)[source]¶ Fit a linear model using ordinary least squares in each voxel.
The design matrix is expected to be 3D because this function is intended to be used in the context of a prewhitened model, where each voxel has a slightly different (whitened) design.
- Parameters
- Yn_tp x n_vox array
Time series for each voxel.
- Xn_tp x n_ev x n_vox array
Design matrix for each voxel.
- Returns
- Bn_vox x n_ev array
Parameter estimates at each voxel.
- SSn_vox array
Model error summary at each voxel.
- XtXinvn_vox x n_ev x n_ev array
The pinv(X’ * X) matrices at each voxel.
- En_tp x n_vox array
Residual time series at each voxel.