Panel Multiple Testing

R and python software for new inference tool in panel multiple testing.

We conduct inference on panels where the number of units and the number of features for each unit are large. Our novel method method:

• provides Family-Wise Error Rate (FWER) control
• allows arbitrary cross-unit covariance
• allows each unit-level model with arbitrary pattern of missing values, feature selections, etc. so their support sets are varying

Codes are in the directories:

• R/funs.R: Inference for panel hypotheses (to be available soon)
• python/funs.py: Inference for panel hypotheses

Installation

There is no need for installation, other than dependent libraries, since our code is fully contained in the two scripts respectively for R and python.

• R: no dependencies
• python: pandas, numpy

R demo

Available soon

python demo

When there are $N$ units and $J$ features, the evidence of unit-level regressions can be stored in a matrix:

• a $P$ matrix $J \times N$ of log p-values;
• whenever $P_{jn}$ is missing, the $j$th feature is not in the support set of $n$th unit-level model.

To run the code, we can select features subject to FWER target of $\alpha$:

import numpy as np
import pandas as pd

J, N = log_pval_matrix.shape
alpha_vec = [0.00001,0.01,0.05] # the FWER thresholds you want to try
pmt_rejection_table =panel_unordered(log_pval_matrix)
rho=pmt_rejection_table['rho'].unique()[0] # the panel cohesiveness coefficient

for alpha in alpha_vec:

	selected_panel_multiple_testing =np.sort(pmt_rejection_table.index[pmt_rejection_table['rho_inv.N.p_1']<=alpha]).tolist()

	selected_Bonferroni_multiple_testing =np.sort(pmt_rejection_table.index[pmt_rejection_table['p_1']<=alpha/(J*N)]).tolist()

Usage

To cite this code, please use

@article{pelger2022inference,
  title={Inference for Large Panel Data with Many Covariates},
  author={Pelger, Markus and Zou, Jiacheng},
  journal={arXiv preprint arXiv:2301.00292},
  year={2022}
}