Geospatial API with Python and PostGIS
Nick Doiron (~mapmeld) |
This workshop empowers you to use geospatial data, with demonstrations of:
- finding free sources of local geodata (official sources plus OpenStreetMap)
- making spatial queries on a PostgreSQL database
- applications for data science / feature engineering
- designing a simple API using your spatial queries
- handling placenames in Tamil, Devanagari, and other scripts
The workshop will be taught with NextJournal notebooks, and a read-only PostGIS server in the cloud, so you can try your own queries right away and after the workshop without installing new software.
Schedule: (10 min) Intro, Preview of our Final Application・ (15 min) Finding and Downloading Local Geodata, Extracting from OpenStreetMap・ (25 min) Intro / Review of SQL and Notebooks・ (20 min) New Types and Functions with PostGIS and SQL・ (15 min) Questions / Break・ (15 min) Importing a CSV and Statistics using PostGIS・ (15 min) Using Statistics in a Feature Engineering / Data Science Context・ (25 min) Intro to Python Flask, Making a Web Service for our Final Application・ (10 min) Considerations for Tamil, Devanagari, and other scripts
The material would be most useful for programmers and data scientists with some previous Python or SQL experience.
There is no mandatory installation step, because the code and PostGIS instance will be in the cloud, connected to forkable notebooks.
These are the setup instructions for a previous, entry-level version of the workshop which I conducted at PyCon Zimbabwe. https://github.com/mapmeld/postgis-zimbabwe This workshop would be freshly updated with Indian geodata and concerns (for example: Unicode / i18n support, flood or disaster management data)
Nick Doiron is a Senior Software Engineer at McKinsey & Company. In the past he has made a variety of open source mapping apps for One Laptop per Child, Code for America, and the Asia Foundation.
Previous Unicode project: https://github.com/mapmeld/wiki-crossword