mosaicperm.panel.MosaicPanelInference

class mosaicperm.panel.MosaicPanelInference(*args, **kwargs)

Mosaic permutation-based inference for linear models in panel data.

Parameters:

*args

Positional arguments passed to the parent MosaicPanelTest class. See MosaicPanelTest for details on required arguments (outcomes, subjects, times, etc.). Note that test_stat is automatically set to None as it is not used in this inference class.

**kwargs

Additional keyword arguments passed to the parent MosaicPanelTest class. Common arguments include cts_covariates, discrete_covariates, clusters, invariance, ntiles, etc.

Notes

For dense covariate matrices, the class uses efficient QR decomposition updates to avoid recomputing regressions for each feature. For sparse matrices, it falls back to recomputing regressions as needed.

Examples

>>> import numpy as np
>>> import pandas as pd
>>> import mosaicperm as mp
>>> 
>>> # Generate synthetic panel data
>>> n_subjects, n_times = 50, 20
>>> n_obs = n_subjects * n_times
>>> subjects = np.repeat(np.arange(n_subjects), n_times)
>>> times = np.tile(np.arange(n_times), n_subjects)
>>> 
>>> # Generate covariates and outcomes
>>> X = np.random.randn(n_obs, 3)
>>> beta_true = np.array([1.0, -0.5, 0.2])
>>> outcomes = X @ beta_true + np.random.randn(n_obs) * 0.5
>>> 
>>> # Fit mosaic panel inference
>>> mpi = mp.panel.MosaicPanelInference(
...     outcomes=outcomes,
...     subjects=subjects,
...     times=times,
...     cts_covariates=X,
...     ntiles=8
... )
>>> mpi.fit(nrand=1000, alpha=0.05)
>>> print(mpi.summary)

The summary will show estimates, standard errors, confidence intervals, and p-values for each coefficient.

Methods

compute_cis([alpha])
compute_mosaic_residuals([verbose])	Computes full mosaic residuals and precomputes useful quantities.
fit([nrand, alpha, features, verbose, ...])	Fits the linear model and computes confidence interva
fit_tseries([nrand, verbose, n_timepoints, ...])	Runs mosaic permutation tests for various windows of the data, producing a time series of p-values.
permute_residuals([_permute_all])	Permutes residuals within tiles.
plot_tseries([time_index, alpha, show_plot])	Plots the results of fit_tseries().
summary_plot([show_plot])	Produces a plot summarizing the results of the test.

Properties

summary

Produces a summary of key inferential results.