This repository introduces techniques for the statistical analysis of spatial data. It covers characterization of spatial data, and techniques for visualizing, exploring and modeling spatial data distributed as point patterns, geostatistical data, and area or lattice data, and methods and problems in spatial data sampling.
- Working with R
- Issues in analyzing spatial data
- General concepts in spatial data analysis
- Methods for spatially continuous data analysis
- Methods for area data analysis
- Methods for point pattern analysis
- Sampling spatial populations
The purpose of this project is to assess the predictive power of kriging (along with a generalized additive model) relative to that of a simple machine learning method like Random Forests, with ordinary least squares serving as the simplest (and most biased) approach. Furthermore, this project is specifically interested in the effect geographic location has on expected earnings for cab drivers in Manhattan. If drivers pick passengers up in downtown Manhattan, they will earn a higher wage on average, everything else equal, implying that geographic latitude is the most relevant factor when determining where to start. This is not a project that seeks to determine which specific variable contributes the most (on average with everything else held equal) to the final fares earned by cab drivers. Rather, it seeks to discover relevant relationship between latitude, longitude and fares collected.