[Statistical inference for regression with imputed binary covariates with application to emotion recognition] - Code and Data Repository

This repository contains the implementation code and research datasets for the paper titled "Statistical inference for regression with imputed binary covariates with application to emotion recognition" published in Annals of Applied Statistics.

Publication Information

Title: Statistical inference for regression with imputed binary covariates with application to emotion recognition

Authors: Ziqian Lin, Danyang Huang, Ziyu Xiong, and Hansheng Wang

Venue: Annals of Applied Statistics (2025)

DOI: 10.1214/24-AOAS1961

Preprint: arXiv:2408.09619

Data Description

data/
├── data_full_final.csv   # The full data
├── pilot_final.csv       # The pilot data
├── W_pilot_final.npy     # The extracted feature vectors
└── README.md             # Detailed data documentation

Code Description

codes/
├── calculate_weight.py  # Compute the optimal weights for weighted estimator
├── cov_impute.py        # Compute the covariance estimator
├── fit_model.py         # Fit the imputed and regression model
├── real_data.ipynb      # Implementation using the demostrated data
├── simulator.py         # Simulators for the algorithm
└── utils.py             # Utility functions

Quick Start

You can run the demo notebook via

jupyter lab codes/real_data.ipynb

Environment

If your environment does not work, try the following version of dependencies:

jupyter-client==7.3.4
jupyter-core==4.10.0
numpy==1.22.4
pandas==1.4.2
scikit-learn==1.1.1
statsmodels==0.14.4

Citation

If you use this code or data in your research, please cite our paper:

APA style:

Lin, Z., Huang, D., Xiong, Z., & Wang, H. (2025). Statistical inference for regression with imputed binary covariates with application to emotion recognition. Annals of Applied Statistics, 19(1), 329-350.

BibTex

@article{lin2025statistical,
  title={Statistical inference for regression with imputed binary covariates with application to emotion recognition},
  author={Lin, Ziqian and Huang, Danyang and Xiong, Ziyu and Wang, Hansheng},
  journal={The Annals of Applied Statistics},
  volume={19},
  number={1},
  pages={329--350},
  year={2025},
  publisher={Institute of Mathematical Statistics}
}

Contact

If you have questions, you can contact the first author Ziqian Lin (linziqian@stu.pku.edu.cn) or the corresponding author Danyang Huang (dyhuang89@126.com)