Arkajyoti Saha

Postdoctoral Scholar, Statistics

WebsiteGoogle Scholar

I am a postdoctoral fellow in the Department of Statistics at the University of Washington hosted by Daniela Witten and Jacob Bien. I received my Ph.D. from the Department of Biostatistics at Johns Hopkins Bloomberg School of Public Health, advised by Nilanjan Chatterjee and Abhirup Datta. I completed my M.Stat (Specialization: Mathematical Statistics and Probability) & B.Stat from Indian Statistical Institute, Kolkata prior to that. My primary research interest lies in developing novel scalable methods for large correlated data. Presently I am working on selective inference approaches for correlated data, where we develop valid statistical inference after correlation thresholding. Another aspect of my research focuses on Spatial Machine Learning where I develop statistical machine learning methods that account for structured correlation in spatial (dependent) data in popular machine learning approaches like Random Forest. I am interested in using these methods to capture spatio-temporal variance in Oceanography data.

The other aspect of my research is Statistical Genetics, where I focus on developing Linkage Disequilibrium (LD)-matrix-based statistical methods for inference on heritability and annotation enrichment which accounts for the correlation across the summary statistics, obtained from GWAS analysis. I have also worked on theoretical development and convergence analysis of Unsupervised Learning methods (clustering) with convex dissimilarity measures & generalized membership functions.