RESEARCH INTERESTS
My primary research interests lie broadly in statistical data science and methodology, motivated by real-life challenges arising from complex systems, social science and public policy. In particular, my doctoral work deals with computational and inferential issues in time-dependent stress-strength interference. In postdoctoral research, I worked on modeling and analysis of incomplete longitudinal data with missingness and zero-inflation. Currently, I am working on causal inference and its application in transport networks. I am also engaged in cross-disciplinary work, focusing on data-analytic settings in sports and environmental sciences.
RESEARCH THEMS
- C.R. Rao
“Statistics is not a discipline like physics, chemistry or biology where we study a subject to solve problems in the same subject. We study statistics with the main aim of solving problems in other disciplines”
Causal Understanding of Transport Interventions
Businesses institutions and federal agencies use machine learning techniques to make sensitive decisions, policy formulation and predictions like whether to grant someone a low-interest loan or estimating how likely interventions in the transport network may reduce accidents. However, if the data used to train an algorithm contains biases against certain characteristics (e.g. races, genders, locations, or other demographic groups), then the resulting estimate will be biased. Using causal methods, we are aiming to ensure algorithmic fairness by taking into account different social biases and compensating for them effectively. In particular, we develop novel inferential methodology to evaluate the causal effects of interventions in London transport networks.
Multiple System Estimation for Hard-to-Count Population
Population size estimation based on a capture–recapture experiment is an interesting problem in various scientific disciplines with a wide range of application areas, for example, the number of civilian casualties in a war or accident, human trafficking, drunk-drivers, and intravenous drug users. Federal agencies are generally interested in such estimates for planning and policy formulation. In general, a census or any registration system often fails to capture all the individuals and that leads to under coverage of the population under consideration. To estimate the under coverage, we propose novel modeling approach to account for the inherent dependence between capture and recapture attempts and heterogeneity in the population.
STUDENTS
2016-17
2017-18
2019-20
2019-20
SERVICE
Stat. Anal. Data. Min., Ann. Appl. Stat.
Member:
Editorial Committee for the ISI blog 'Statisticians React to the News'
Review Panel of the Statisticians for Society initiative by the RSS
Organizer: Invited Paper Sessions, 63rd ISI WSC 2021
Advancements in Capture-Recapture Methods with Application in Social sciences and Humanitarian Crisis
Recent Advancements in Causal Analysis with Application in Policy Decisions
SELECTED PRESENTATIONS
Analysing causal effect of London cycle superhighways on traffic congestion.
RSS International Conference, Manchester, September, 2021
2019 International Indian Statistical Association Conference, Mumbai, December, 2019
European Meeting of Statisticians, Palermo, July, 2019
Estimation of reliability with semi-parametric modeling of degradation.
11th International Conference on Mathematical Methods in Reliability, Hong Kong, June, 2019
22nd International Conference on Computational Statistics, Oviedo, August, 2016
A Bayesian Two-stage Regression Approach of Analyzing Longitudinal Outcomes with Endogeneity and Incompleteness.
ISBA World Meeting, Edinburgh, June, 2018
2017 International Indian Statistical Association Conference, Hyderabad, December, 2017
Estimation of reliability with cumulative stress and strength degradation.
2015 International Indian Statistical Association Conference, Pune, December, 2015
Identifiability issues in dynamic stress-strength modeling.
Eighth International Workshop on Simulation, Vienna, September, 2015
Dynamic stress-strength modeling with cumulative stress and strength degradation.
21st International Conference on Computational Statistics, Geneva, August, 2014