pymare.stats.weighted_least_squares

weighted_least_squares(y, v, X, tau2=0.0, return_cov=False)[source]

Perform 2-D weighted least squares.

Parameters:

  • y (numpy.ndarray) – 2-d array of estimates (studies x parallel datasets)

  • v (numpy.ndarray) – 2-d array of sampling variances

  • X (numpy.ndarray) – Fixed effect design matrix

  • tau2 (float, optional) – tau^2 estimate to use for weights. Default = 0.

  • return_cov (bool, optional) – Whether or not to return the inverse cov matrix. Default = False.

Returns:

If return_cov is True, returns both fixed parameter estimates and the inverse covariance matrix; if False, only the parameter estimates.

Return type:

params[, cov]