Hacker News new | ask | show | jobs
Show HN: Visualizing 9 years of relationships with instant message frequency (hipsterdatascience.com)
11 points by cbabraham 3815 days ago
3 comments

Visualizing this dataset has been my end of year side project for a few years now.

The hardest part was matching hundreds of friend's various usernames across different services to the same person. It's a process that's difficult to automate, even with some helper scripts it took a couple days to manually curate the mapping.

The svgs for the grid are in pre-rendered static files made with node/jsdom, which was pretty fun to write.

Data sources:

Skype history file

Facebook data download

Google Voice and Google Hangouts data download

Email (for oldschool gchats)

iPhone backups, for SMS/iMessage

Slack API

I have also been collecting my sms data for almost a decade now.

Symbian and Android had options to extract to xml/txt. But after I got Windows Phone and iPhone, it is harder for me to extract all texts but they are there in the cloud. I will have to extract and normalize them somehow.

Super impressive that you kept all that data over so many years. Great job!