Copy
Thanks for reading my newsletter. Special shoutout to the new subscribers! Read on for new learning resources and updates from the world of free and open-source geospatial tech.

Upcoming Classes


Advanced QGIS Certification, 4-5 September, 2020, at EMEA/AMER-friendly time
Few seats left to join my live online class this week and earn your official QGIS.org certification. Register now!
My Advanced QGIS Class from August 2020. 10 participants from 6 countries, with diverse backgrounds in geology, urban planning, hydrology, environmental sciences, agriculture, demography - all united by QGIS!

Featured Topic: Spatial Data Science
 

I have been spending a lot of time researching and developing my new course Spatial Data Science. This is an important topic but the current offerings do not do the justice to the 'spatial' aspect. Either these are GIS courses repackaged with a shiny new name or are too simplistic that does not leverage the full potential of spatial datasets.

Most of the attention has been on Deep Learning methods - object detection and extracting features from imagery. While this is useful, I see a huge opportunity to apply simpler statistical and machine learning techniques to problems in the spatial domain that remains unexplored.

So that's where I am focusing my efforts. Developing a comprehensive introductory course on machine learning that leverages spatial datasets and spatial methods to solve real-world problems. I wanted to give you a sneak peak and share some thoughts.
  • A module in the course focuses on using Uber Movement and OpenStreetMap data to build a model that can predict travel times in a city taking into account traffic patterns. As the tools have become so good, with just a little bit of code, you can get results that can match what Google Maps can do. See my recent submission to PyCon India that was the result of developing this module.
  • Another module explores regression methods to work with precipitation data. We start by a simple model and gradually add more parameters (elevation, temperature etc.) to show how we can predict rainfall at any given location. We then use geostatistical methods to account of spatial autocorrelation and learn how to implement it in Python.
  • My favorite exercise is from the module on urban change detection. We take public domain satellite imagery and with very little training data, create a robust model that shows bare land that has changed to built-up area - showing construction activity. Here's an interactive app showing this in action. We implement this model in Python and then use this to predict urban growth.
There is much more in the works and hope to offer this class soon. If you have suggestions on other problems to tackle or pointers at material - please let me know.
I experienced what it feels like to go viral on Twitter after announcing my free self-study courses :)  I heard from hundreds of people and thousands have downloaded these courses since the launch. I am humbled by such huge interest and validates my mission to create high-quality learning resources that are accessible and affordable. Visit Spatial Thoughts OpenCourseWare site to see all the courses.



 

Learning Resources and News

  • Lutra Consulting kicked off a crowdfunding campaign for Point cloud data support in QGIS. If this campaign is successful, QGIS will be able to read, view and style LIDAR data natively. I pledged to this campaign and encourage you do contribute too!
  • University of Wisconsin's Cartography Lab released a complete book on webmapping using open-source technologies. Check out the full book online The Web Mapping: A Workbook for Interactive Cartography and Visualization on the Open Web
  • In my quest for good machine learning materials for spatial data, I discovered this excellent training by Madlene Nussbaum - covering the theory in a lot of depth. There are 2 video lectures [1] [2] accompanied by exercises
  • Dan Mahr has put together a problem-based guide of common CRS issues, root causes, and solutions. If you find yourself frustrated with CRS issues, head over to https://ihatecoordinatesystems.com
You can browse, search and filter current and past learning resources in this public spreadsheet.
Lastly, Spatial Thoughts team is growing! Santhosh joins me today as a Training Associate. He will work with me to accelerate creation new tutorials and open course materials. You'll hear more about his work in future newsletters.
 
Regards, and Stay safe!

-Ujaval
LinkedIn
Twitter
Website
Copyright © 2020 Spatial Thoughts, All rights reserved.


Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.