gemmr.sample_size.linear_model.do_fit_lm

gemmr.sample_size.linear_model.do_fit_lm(ds, n_reqs, include_pc_var_decay_constants=True, include_latent_explained_vars=True, include_pdiff=False, verbose=False, prefix='')

Fits a linear model to outcome data.

First, prepares outcome data, i.e. selects predictor variables and stacks it along dimensions px, r, Sigma_id. Second, fits linear model.

Parameters:

  • ds (xr.Dataset) – outcome dataset

  • n_reqs (xr.DataArray) – required sample sizes

  • include_pc_var_decay_constants (bool) – whether to include a predictor for the principal component spectrum decay constant in the linear model

  • include_latent_explained_vars (bool) – whether to include a predictor for the latent explained variance in the linear model

  • include_pdiff (bool) – whether to include predictor for \(|p_X - p_Y|\) in the linear model

  • verbose (bool) – if True prints deltaAIC

  • prefix (str) – prefix for outcome variables in ds

Returns:

  • lm (sklearn.LinearRegression instance) – fitted model

  • X ((n_synth_datasets, n_predictors)) – predictor data matrix

  • y ((n_synth_datasets,)) – dependent variable

  • coef_names (list) – labels for included linear model coefficients (first one is “const”)