Last summer, I started a project on visualizing topic clusters from TED Talks 2013 - that can be seen in some of previous posts. I felt that the tool I had used to display my topic clusters was, well, a little clunky. I can't really complain because Datacratic open sourced it and they inspired me to make ....

dreda

which is an acronym of this posts title (& for my linguist friends, its dreɪ'ɾə ).

If you check it out, you'll see there are a few examples of data to choose from - the first one is just the data from my previous posts on TED Talks and really the impetus for the project - just looking at data from different angles.

The other examples are the infamous fisher's iris flowers data set, some open data from SF OpenData, and the dataset from the datacratic post that inspired me to look at this from the beginning.

If you decided to look at the tutorial, you'll see the structure needed to upload your own file for visualization:

The project as a whole was really a chance for me to check out three.js and the possibility of doing more data visualization in the browser. I've used d3 before but was curious how three.js would hold up - its definitely a lot of work, but I enjoy the product. It was also a fun experience getting to know semantic-ui a little better through this project. Also, learning how to fully crash browsers before I found out what billboarding was - thanks to Winnie Wang <3.

For me, the end goal was met - having a place to immediately visualize dimensionality reduced datasets from an iPython notebook in a fun, interactive modality.

Of course you'll find some bugs in it - but I open sourced it so complaints = pull requests, right? Haha, but seriously, check out the source here if you're interested.