Making awesome apps with Open Data - Erik Romijn¶

Tags: python, pun

(Summary of a talk at the Dutch Python meeting)

Note: this is about any kind of app, not specifically python or mobile. Erik worked a lot on open data apps, for instance for Amsterdam (http://www.appsforamsterdam.nl/).

Our governments have tons of open data, but they often don’t even know they have it: they have to search for it. They often publish the data for free, usually CC0 or CC-BY. Often it is more open than wikipedia or openstreetmap. Water level data. Data about Amsterdam fire truck depatchings. Registry of valuable trees in Nijmegen. Look at http://data.overheid.nl/ for a list of what’s available in the Netherlands.

Open data is crude oil to an App Maker—money just sitting on the ground waiting for someone to suck it up. (link)

You can enrich your app with info about the world around it. You can turn data into knowledge. Lots of data is pretty raw, which means you can do a lot with it. We, as programmers, as people that can do something with this data basically have superman powers. We can do something with it, most people can not.

There are four keys to making a wonderful open data app:

• Awesome visualizations. Perhaps simple and accessible. If you visualize water levels in an easy way, people will look at it when there’s high water.

• Innovate with the dataset. Combine multiple datasets. For instance http://hogenood.nu/ where you can find the closest public toilet? With data taken from lots of data sets.

• Make the data exciting, even though the original data might be boring (“energy ratings for buildings…”). Energy ratings for buildings are boring until you can compare your score with those of your neighbours!

• Does your app improve social relevance? http://10000scholen.nl/ compares school ratings and scores so you can pick a better school for your kids.

If you’re thinking about building one, look at open data contests. They’re free to participate in. Bonus: there are often prizes, ranging from 100 to 3000 Euro. Also there are hack-a-thons for open data in the Netherlands. They’re a great place to get started. And a great place to find data, as the data owners are often also there.

Open data is not without its challenges, though. Some data sets are really incomprehensible without domain knowledge: it is just too hard to understand. Data formats can be hard. In any case, data comes in lots of different formats: csv, json, xml, shapefiles. Access through soap, http, ftp, usb sticks. Geographical data in the Netherlands are often in rijksdriehoek coordinates instead of the web’s google mercator coordinates. And… the http://data.overheid.nl is all in Dutch and they’re not planning an English version yet.

After you’ve made your app, make sure it is durable. That it doesn’t stop working. And talk to the data owners! Tell them what you did!

Where from here? There are more contents planned. And we’re passing the stage where boring is OK: apps have to be really good now. And, as our apps become more awesome, we will collectively get more and better data. If open data is used in a beautiful and attractive way, more people will want to release their data as open data.