Contributions to the Total Good
Head over to hobsonlane.com for a slightly better experience
This totalgood.com django-based blog app I built is pretty horrible. To keep it db-free (and compatible with github.io and free-tiers at heroku and aws), I manually edit dicts in my views.py to create content. The new Jekyll webpage templating platform (which is also database-free) seems pretty nice. github.io uses it by default to compile their `gh-pages` branches, so I figured I'd give it a shot and reduce my dependence on heroku. But I won't be able to do anything dynamic (data-ish) over there. Steven Skoczen has come out with a really slick django-based blogging app. So I'll be firing that up soon (to add to my list of unpolished/unmaintained blogs). But, for now, the only place I've got anything decent to look at is at hobsonlane.com.
Force-Directed Graph (Network) Visualization
Simple "bag of words" PCA on the US Presidential innaugural speeches. The word/token occurrence counts in each document are used to determine the stiffness of springs in a D3.js force-directed graph as well as "rest" length of the springs. The larger the number of shared words between documents, the shorter the "rest length" and stiffer the spring connecting those two documents (graph nodes). See my PDX Python presentation at www.hobsonlane.com/pug/ if you want to learn more.
D3.js Line and Bar Charts
I've written some pythong functions Takes lists of lists (e.g. table) and plots them 2 ways. Unfortunately, I'm having trouble deploying all the dependencies of the `pug` package to heroku (numpy, scipy, pandas, and nltk are the tough ones), so the demo dashboard at www.storydecoder.com/dashboard/ with pretty plots is usually broken.
Stock Price Prediction + Trading Algorithm Backtester
Here are some of the fun things I learned at Tucker Balch's Computational Investing class on coursera. I'll share some event studies and trading algorithms as soon as I figure out how to share ipython notebooks within a Django app