Prototype Wrapup #2
Week 2 of my prototype adventure fell off the rails a little. Last week I committed to 3 prototypes, but I only made one:
Relative word cloud
This is an idea I wanted to try out. Traditional word clouds are really distorted in favour of common words. Normally you filter the most common out with stoplists, but those are a really blunt tool. I thought it would be more fun to use a base set of words and do some Bayesian magic to make a word cloud showing the words that appear a lot in the sample text, but not much in the base set. Unfortunately, the statistics involved was kind of out of my league and I ended up spending way too much time twiddling numbers trying to get something I was happy with. That said, I am pretty happy with it.
Time: 12 hours.
I suppose it shouldn't be too surprising that I didn't get any others done given the amount of time that I sunk into this one. However, I also really underestimated how much time the Christmas period would take up, so I think I could have still achieved the 3 if I'd planned things a bit better. I'm going to commit to 3 more for next week, and keep trying to get the time-per-prototype under control.