A few notes from this week:
- There's a huge debate on the Frequentist vs Bayesian schools of thought with proponents on both sides. Just in case you're on the fence here's an Open Letter on why you should think about going Bayesian
- The goal in Bayesian inference is to get a good handle on the posterior distribution over the input parameters. Some of the math can get pretty beefy so you would normally use a package like pymc or rjags to fit you distributions / models. Some good resources for pymc and MCMC are this Stats Book for Hackers and this set of videos from mathematicalmonk
- Some EDA (Exploratory Data Analysis) tools you should have in your workflow include Raw and CartoDB
- We had two pretty good talks by @nitin on LearnDataScience and @Udacity on their Data Science course development at one of the local meetup groups
wonderful....
ReplyDelete