Feature engineering for spatial data - Alexander Meshcheryakov

Dive into the realm of spatial data analysis with a talk by Alexander Meshcheryakov, a trailblazer in the field. Join us for an illuminating session where Alexander delves into the art and science of feature engineering tailored specifically for spatial data. In this talk, you'll discover how to extract meaningful insights from geographic information and unleash the true potential of location-based datasets.

00:00 - Introduction
01:35 - How to get geodata?
02:22 - Geometry representations in WKT
03:11 - Geocoding
03:43 - Latitude and longitude as a feature of the model
04:16 - Polygon conversion - buffer
04:48 - Polygon transformation - Convex Hull
05:17 - Polygon features
06:14 - Tobler's first law of geography
08:19 - Spatial joins
08:38 - Why else would we need spatial joins?
09:41 - How to get a point from a polygon
10:09 - Distance calculation: Planar distances
10:56 - Problem of Earth projection
11:30 - Mercator projection
12:14 - Variety of projections
12:40 - What is SRID?
13:13 - Transition from one projection to another
13:37 - Geodetic distances: haversine distance
14:14 - Earth Model WGS 84
14:45 - Route distance
15:43 - Spatial data aggregation methods
17:08 - Number of adjacent points in ring
18:01 - Search of hotspots and outliners
18:58 - John Snow's map
20:40 - Clustering
21:06 - KDE - kernel density estimation

#python #spatialdata