Brief about myself
Welcome to my webpage! Currently, I am a postdoctoral associate in the Department of Biostatistics and Bioinformatics at Duke University working under the mentorship of Sheng Luo. Here is my most recently updated CV.
I received Ph.D. in Statistics from North Carolina State University in 2021. My dissertation work primarily focused on developing inference techniques for functional data. I completed Masters of Statistics from Indian Statistical Institute, Kolkata in 2015. The entire credit of sparking my interest in the subject of statistics goes to Ramakrishna Mission Residential College, Narendrapur, Kolkata, where I received the Bachelors of Science degree in 2013.
My research is motivated by the quest of answering scientific questions driven by the data that can be observed as trajectory, commonly attributed as functional data in the nomenclature of statistics. Depending on the sampling design, functional data is broadly divided into two categories, i) Dense functional data: the data of this genre comes from domains from but not limited to medical imaging, wearable devices, ii) Sparse functional data: this data occurs when the observations are observed sparsely for each subject, data observed under a longitudinal design is often encompassed under the sparse category.
From a statistical perspective, many of my research works concentrate on developing valid inference procedures for functional data observed over a rather complex design, such multivariate functional data or longitudinal functional data. Such type of data are becoming common everyday, and they are gathered together under the framework of second-generation functional data. Have a look at my recent review on such kind of data published in Annual Review of Statistics.
Apart from functional data techniques, I am also interested to work on topics related to foundations to statistics such as Generalized fiducial inference (GFI), conformal predictions and apply them a toolbox to statistical learning. As a part of my dissertation work, I have developed an Epsilon-admissible subset (EAS) framework for variable selection in multivariate linear regression setting using a GFI approach. See the recently published article here.
Current research interests
Statistics on second-generation functional/shape-object data : Develop projection-based tests for complex hierarchically structured functional or share object data, such as longitudinal functional data, multivariate functional data, and with an extended focus on inference from observational studies.
High-dimensional statistical inference: Research on Epsilon-admissible set methodology for variable selection in high-dimensional data.
Machine learning on intensively collected digital bio-signatures data: Inference methods for electronic health records and wearable device data for Alzheimer’s and Parkinson’s disease.
Analysis of longitudinally recorded clinical and brain imaging data: Statistical methods to integrate longitudinal brain imaging data with clinical data for early detection of Alzheimer’s disease.
Foundations of Statistics: Research on application of generalized fiducial inference on non-standard statistical settings, and comparison with Bayesian and frequentist regimes.
Construction of any-time valid prediction sets: Research on constructing distribution-free conformal prediction sets for complex mixed-type longitudinal functional data.
https://orcid.org/0000-0003-1952-4210