My research focuses on developing statistical methods to address fundamental data challenges, including heterogeneity, measurement errors, missingness, and zero-inflation. These challenges arise in diverse domains such as genomics, epidemiology, and electronic health records, where conventional statistical models often fall short. To tackle these issues, I leverage and extend methodologies in quantile regression, machine learning, and debiasing techniques from measurement error analysis. My goal is to create robust, interpretable, and computationally efficient approaches that improve inference and prediction in complex data environments.

Understanding Heterogeneous Data with Quantile Regression

Real-world data often exhibit complex, nonlinear, and heterogeneous relationships that traditional mean-based methods fail to capture. I develop and extend quantile regression techniques to model variability across different parts of the data distribution, enabling more insights into heterogeneous associations. This is particularly relevant in applications such as personalized medicine and socioeconomic studies, where responses vary across subpopulations.

Correcting Bias in Complex Data with Measurement Errors

Many datasets contain errors due to mismeasurement, rounding, or systematic bias, particularly in high-dimensional, count, and compositional data. I work on error-in-variables models, de-biasing techniques, and correction methods that enhance the reliability of statistical inference. My approaches improve estimation and hypothesis testing in settings like nutritional epidemiology and biomedical research, where measurement errors can significantly distort conclusions.

Advancing Learning Methods for Missing and Zero-Inflated Data

Missing values and zero inflation are prevalent in fields such as clinical trials, microbiome studies, and financial data. I develop statistical and machine learning approaches, including transfer learning and imputation techniques, to mitigate bias and improve predictive performance. By designing methods that adapt to data sparsity and structural zeros, I aim to enhance inference in cases where traditional methods struggle with loss of information.

  • Zhao, H., and Wang, T.✉ (2025+). “Generalizing Transfer Learning: A Flexible Doubly Robust Estimation Approach for Missing Data”, under review.
  • Zhao, H., and Wang, T.✉ (2025+). “Doubly robust augmented model transfer inference with completely missing covariates”, under review.
  • Wang, Z., and Wang, T.✉ (2024). “A Semiparametric Quantile Single-Index Model for Zero-Inflated Outcomes”, Statistica Sinica, accepted.
  • Wang, T.✉, Zhang, W., and Wei, Y. (2024). “ZIKQ: An innovative centile chart method for utilizing natural history data in rare disease clinical development”, Statistica Sinica, accepted.
  • Wang, Z., Ling, W. and Wang, T.✉ (2024). “A Semiparametric Quantile Regression Rank Score Test for Zero-inflated Data”, under review.

Other topics

I am also interested in a broad range of research topics, such as high-dimensional statistics and case-control studies.

  • Wang, Y., and Wang, T.✉ (2025+). “Multi-Group Quadratic Discriminant Analysis via Projection”, under review.
  • Wang, T., Liu, J., and Wu, A. (2024). “Semiparametric Analysis in Case-Control Studies for Gene-Environment Independent Models: Bibliographical Connections and Extensions”, Journal of Data Science, accepted.
  • Ma, S. and Wang, T.✉ (2023). “The optimal pre-post allocation for randomized clinical trials”. BMC Medical Research Methodology, 23:72 doi: 10.1186/s12874-023-01893-w.
  • Wang, T.✉ and Asher, A. (2021). “Improved Semiparametric Analysis of Polygenic Gene-Environment Interactions in Case-Control Studies”. Statistics in Biosciences, 13, 386–401.
  • Gaynanova, I. and Wang, T. (2019). “Sparse quadratic classification rules via linear dimension reduction”. Journal of Multivariate Analysis, 169, 278–299.

underline indicates a student working under my (co)supervision, with ♦ denoting an undergraduate student mentee; ✉ indicates the corresponding author.

Research opportunities are open to highly motivated students. Interested individuals are encouraged to reach out for more details.