"Data is like garbage. You'd better know what you're going to do with it before you collect it."
About Me
I am a PhD Candidate in Statistics in the Department of Mathematical Sciences at Michigan Technological University, Houghton, Michigan, where I work under the supervision of Prof. Xiao Zhang. My research focuses on Bayesian modeling for discrete and categorical data, with particular emphasis on MCMC methods, missing data, and the Multinomial Probit Model (MNP).
My current work develops parameter-expanded data-augmentation methods for binary, ordinal, and nominal outcomes with missing values. This research has been presented at leading statistical conferences including JSM 2024, JSM 2025, and SDSS 2024. I was also honored with the Dean’s Award for Outstanding Scholarship (Spring 2026) at Michigan Tech.
Before starting my doctoral studies, I worked as a data analyst and statistician on large-scale survey and policy-oriented projects supported by organizations such as the World Bank, NORC, and other international partners. My work combines methodological research with applied statistics, survey analysis, and real-world problem solving.
Research Interests
- Bayesian Computation for Latent Variable Models
- MCMC Methods and Parameter Expansion
- Missing Data Imputation
- Multinomial Probit and Categorical Data Models
- Multivariate and Longitudinal Data Analysis
- Applied Survey Statistics and Biostatistics
Current Work
I am currently developing Bayesian MCMC methods for binary, ordinal, and nominal measures with missing values using the Multinomial Probit Model (MNP). My recent work includes:
- Developing a parameter-expanded data-augmentation Gibbs sampler to improve posterior mixing and convergence
- Extending MNP models to multivariate and longitudinal settings
- Integrating missing-data handling directly within the Bayesian estimation framework
- Evaluating methods through simulation studies and real-data applications
Experience
Graduate Research Assistant — Michigan Technological University, Houghton, MI (Jan 2021 – Present)
Research on Bayesian modeling for nominal and categorical outcomes with missing values using the Multinomial Probit framework; development of efficient MCMC algorithms and parameter-expanded Gibbs samplers.
Graduate Teaching Instructor — Michigan Technological University, Houghton, MI (Aug 2021 – Dec 2023)
Instructor for Undergraduate Engineering Statistics and Precalculus; taught labs using R and MATLAB; designed assignments, quizzes, and course materials.
Data Analyst / Statistician — Solutions Consultant, Kathmandu, Nepal (Dec 2017 – Jul 2021)
Led statistical analysis, sampling design, questionnaire development, and reporting for large survey-based and policy-focused projects funded by international organizations.
Teaching Instructor (Part-time) — Amrit Campus, Kathmandu, Nepal (Dec 2017 – Jul 2021)
Taught Survey Statistics, Demography, Biostatistics, and Research Methodology; supervised student projects and practical sessions.
Selected Publications
Silwal, S., & Zhang, X. (2025). Parameter-Expanded Data Augmentation for Analyzing Discrete Nominal Measures with Missing Values Using Multinomial Probit Models. Submitted.
Silwal, S., & Uprety, P. (2019). Assessment of Oral Health Knowledge, Attitude and Practice Among School Children in Kathmandu Metropolitan City, Nepal. RRJoST, 8, 1–10.
For a full list, please see my Publications page.
Recent News
Quick Links
Contact
I am interested in research collaborations in Bayesian statistics, categorical data analysis, missing data methods, and applied statistical modeling. Please feel free to contact me at suwashs@mtu.edu.