• No perfect projection: choose the one that minimizes distortions in shape, direction, size – whichever is more important for your task. Remember several core projections for our area. Remember EPSG:4326. Think of how you are going to use the data: operations, accuracy requirements, compatibility, the cost of reprojection.
• Be explicit about projections when performing spatial operations in Geopandas or plotting several GDFs together (or changing geometry in GDF). Avoid "naive geometries".
• Understand how to compute distances. Use projected CRS to measure distances and areas - not decimal degrees!
• There are conventions in GIS about core concepts such as distance between spatial features (“shortest separation between features”; not between their centers), and equality of spatial objects, which is defined through their geometric properties (such as properties of exterior and interior rings in case of polygons) and not through a list of coordinate pairs: note this when joining data
• Topological operations (such as adjacency) won’t always give you a good answer if geometries aren’t correct. This can happen for many reasons.
• Constructive (understand Buffer, Centroid, Representative_point) and set-theoretical spatial operations return GeoSeries. You may need to convert it to GeoDataframe (e.g., for Spatial Join)
• GeoDataframe can be constructed using a Point constructor to create GeoSeries; from a pandas Dataframe; from a GeoJSON object
• Spatial Join - joining two GeoDataframes based on spatial relationship between geometries
• Overlay: similar to spatial join, but creates new spatial features (as a new GeoDataframe)