Sitemap
A list of all the posts and pages found on the site. For you robots out there, there is an XML version available for digesting as well.
Pages
Professional Development
Honors, certifications, skills, and coursework at CMU and UNT.
Posts
Could offence be the best defence in football?
Published:
A review of “parking the bus” at the 2026 World Cup, combined with data on whether attacking teams are better at protecting a lead.
“For 60 overs, they should feel like hell out there.”
Published:
How much of India’s 2021 Lord’s turnaround can actually be attributed to Kohli’s captaincy? Expected runs/wickets models plus empirical Bayes shrinkage estimate the persistent effect of Test captains.
projects
Allegheny Land Trust: St. John’s Citizen Science Analysis
A data-for-good project with CMU Students Using Data for Good, measuring whether citizen-science activity around the Allegheny Land Trust’s St. John’s project has grown faster than state and county baselines. Download the write-up (PDF)
CUSUM Portal: Spatio-Temporal Earthquake Anomaly Detection
An in-progress anomaly-detection portal that flags abnormal seismic swarms in live USGS earthquake data using spatio-temporal CUSUM change detection. Research with Dr. Woody Zhu, supported by SURF.
Geospatial Mobile Application for Emergency Response and Management
A location-based AR app for navigation and emergency response on the UNT campus, built on Google Photorealistic 3D Tiles and Cesium for Unity. Research in UNT’s Data Visualization and eXtended Reality (DVXR) Lab.
NFL Big Data Bowl 2026: Ball Hawks
Measuring a defender’s commitment to the ball when it is in the air and its impact on play outcomes. Submitted to the NFL Big Data Bowl on Kaggle.
Syracuse Basketball Analytics Competition
Built a composite metric using weighted min-max normalization across Expected Points, Advantage Ratios, and Points Per Chance Ratios to identify which players a new team should prioritize protecting or targeting from proprietary data metrics.
TAMS Mock Election 2024
A student-run mock election at the Texas Academy of Mathematics and Science: 115 voters, Texas-style ballot ordering, and a look at what ranked-choice voting can and can’t fix. Download the results write-up (PDF)
publications
Modeling American Kestrel Decline Using Spatiotemporal Subsampling to Improve eBird Data Reliability
Published in bioRxiv (preprint), 2025
Debiases eBird citizen-science data with spatiotemporal subsampling, then uses a time-series model to test whether rising Cooper’s Hawk numbers help explain the American Kestrel decline.
Recommended citation: Sharma, A., Bednarz, J. C., & Kwon, J. (2025). "Modeling American Kestrel Decline Using Spatiotemporal Subsampling to Improve eBird Data Reliability." bioRxiv. doi:10.1101/2024.10.31.621376.
Download Paper
Stats & Stumps: Using Machine Learning to Predict T20I Matches with Player and Venue Data
Published in Wharton Sports Analytics Journal, 2025
Predicts Twenty20 International outcomes from venue-adjusted, player-level impact metrics. Regularized logistic regression reached 70.24% test accuracy on 1,029 matches.
Recommended citation: Sharma, A. (2025). "Stats & Stumps: Using Machine Learning to Predict T20I Matches with Player and Venue Data." Wharton Sports Analytics Journal.
Download Paper
talks
Luck can be a Funny Thing Permalink
The lessons we can take from statistics to our lives.
Relationship lessons from the stock market Permalink
Published:
The similarities between investing in the stock market and success in human relationships, presented at TEDxTAMS 2024.