Remote Sensing and Geospatial Data Analytics
Toledo, OH
I am a geospatial professional with 3+ years of experience in solving diverse sets of problems. Currently, I work as a Research Scientist, specializing in data science applications for remote sensing. In my day today, I develop statistical models for tree species classification, use Py6S for radiation calibration or perform a literature review on GHG emissions, to name a few. I have held various geospatial data-centric positions where my responsibilities varied from physical data collection to curation to modeling.
ACADEMIC WORK
Worked on grid-based analysis using Spatial Analyst, automated techniques using ArcMap, Netogo simulation techniques and R Studio. Specializing in environmental modeling decisions and considerations. Special attention to strong environmental data visualization and using map design.
Implemented transportation modeling in GIS environment. Designed travel demand modeling and logistic analysis. Explored the application of GIS in emerging data and tools in the field of transportation analytics.
Focused on theories and implementation of spatial networks with an emphasis on non-planar networks especially social network data.
A mixture of GIS and statistics focussing on node properties, diffusion, communities, and network configurations, overlaid and fused with other spatial data.
FOCUSSED PROJECTS
WHAT PARTS OF MANHATTAN ARE MORE CONNECTED THAN OTHERS?
Spatial Network
-
Developed modularity clusters of New York city to understand the strength of connection in the city. Modularity classes use cab rides within the city at each timestep. Timesteps were decided based on travel density.
-
Devised clusters were spatially contiguous and highly dependent on landuse and socioeconomic status of the area.
PREDICTIVE ANALYSIS OF THE IMPACT OF WEATHER ON TRAVEL TIME
User Interactive Tool
-
Investigated spatial-temporal patterns of 2 million cab rides in NYC through linear regression and supervised learning
-
Optimized performance of the regression model through VIF, ANOVA, and regularization enhancing R-squared by 40%
-
Benchmarked performances of SVM, XGBOOST, and Random Forest models, achieving the best accuracy of 72%
WHERE IS THE VIEW?
User Interactive Tool
-
Created a standalone python script that lets you find location based on your selected view feature.
-
Executed visibility analysis using stratified systematically sampled points at the highest 90th percentile elevations. Overlayed output to find the selected view feature. Outputs the geo-coordinates of location with the view.
FLOOD RISK MAPPING USING REALTIME WATER LEVEL: STUDY OF FLORIDA
User Interactive Tool
-
Compiled a user interactive python tool to delineate flood risk zones using various influence factors.
-
Developed a standalone python program to interpolate and classify current water levels with option to compare different interpolation techniques.
-
Delineated areas under high risk. (high flood risk and high water level)
