Biodiversity Informatics | Computational Taxonomy | Spatial Data Science
I am a researcher specializing in quantitative ecology and biodiversity informatics. My work focuses on bridging the gap between computational data science and ecological theory to address severe data gaps and biases in the Global South.
I build reproducible machine learning workflows to analyze large-scale open biodiversity datasets, audit data quality, and model species distributions.
- Programming & Statistics: R (dplyr, terra, sf, ggplot2, ranger, mgcv), Python, SQL
- Remote Sensing & GIS: QGIS, Google Earth Engine (GEE), Sentinel-2, MODIS
- Methods: Species Distribution Models (SDMs), Spatial Bias Assessment, Machine Learning (Random Forest, XGBoost, GLMs)
- biasProxyR (Published on CRAN): An R package that addresses geographic sampling bias in Species Distribution Models for data-poor regions. It generates sampling effort proxies using open-source infrastructure data to prevent models from learning human sampling artifacts.
- taxaudit: A publication-grade R package for ontology-mapped botanical trait extraction, agreement auditing, and leakage-free evaluation.
- rOpenSci Contributions: Actively contributed to core biodiversity infrastructure, including bug fixes in
taxize, documentation improvements inrgbif, and reproducible workflows inspocc.
- Wildfire Recovery Geospatial Modeling: Spatially explicit modeling of post-fire vegetation recovery across 15 California wildfires (2018โ2022). Uses Sentinel-2, terrain, and climate data to evaluate ecosystem resilience and transitions.
- GBIF Data Quality Pakistan: Automated data quality audit of 195,453 GBIF plant records, flagging severe spatial errors and centroid clustering.
- GBIF Spatial Bias Pakistan: A quantitative analysis demonstrating that 75% of national records originate from a single province, highlighting severe geographic data concentration.
- Khejri Ensemble SDM Spatial CV: Spatially-validated ensemble species distribution modeling for Prosopis cineraria, proving the necessity of spatial block validation over random splitting.
- Cholistan Vegetation Stress: Machine learning framework for detecting seasonal vegetation anomalies in the Cholistan Desert using MODIS NDVI and climate data.
- ResearchGate: Aamna Naveed
- Email: naveedaamna4@gmail.com