Talk Proposal Submission

If you are interested in attending this talk at PyCon JP 2017, please use the social media share buttons below. We will consider the popularity of the proposals when making our selection.


Geospatial data analysis and visualization in Python(en)


Halfdan Rump

Audience level:



Useful libraries


In this talk I will introduce you to some very useful libraries for geospatial data visualization and analysis. I will show you how to create your own maps and how I solved the problems that I ran into. I will use data from 食べログ and SafeCast. If you are interested in data mining, visualization and, of course maps, then this talk is for you.


In this talk you will learn how to create great looking interactive maps for visualizing datasets with geospatial coordinates. I'll show you what Python libraries make it easy to create such maps. My aim is that you will leave the talk with a desire and knowhow to start making interactive maps using Python on your own.


Interactive maps are great for exploring and getting a quick intuition of datasets containing location information. In this talk I will show that you don't have to be a data scientist or a JavaScript expert to create such maps. More concretely, I will give a quick intro to some great libraries and show you how to: - use `osmnx` to download map data and convert it into a street graph - use `geopandas`, `networkx` and `shapely` to manipulate street graphs and assign data points to areas - use `pyproj` and `geopy` for changing between coordinate reference systems and measuring distances (I'll give you a short demonstration of how important this can be) - use `folium` for creating beautiful and responsive maps that are rendered to HTML and JavaScript This covers the basic part of the talk, and I will then move into the second part, talking about some of the more difficult issues that I encountered while creating maps: - how to deal with lack of geojson/shapefile boundary data for small areas - how to deal with geospatial data that changes over time Apart from the tools mentioned above, I'll show you how `networkx`, `scikit-learn`, and good old plain Python can be used to solve these problems. 英語で発表しますが、質問は日本語で受け取ります。
  • このエントリーをはてなブックマークに追加