The course emphasizes how to design A/B tests using prior “guestimates” of effect sizes (often you have these from prior campaigns, or somebody claims an effect size and it is merely your job to confirm it). It is fairly technical, and the emphasis is Bayesian- where we are trying to get an actual estimate of the distribution unknown true expected payoff rate of the various campaigns (the so-called posteriors). We show how to design and evaluate a sales campaigns for a product at two different price points.
Win-Vector LLC is starting a data science mailing list that we would like you to sign up for. It is going to be a (deliberately infrequent) set of updates including Win-Vector LLC notices, upcoming speaking events, and data science products.
One of the big points of Practical Data Science with R is to supply a large number of fully worked examples. Our intent has always been for readers to read the book, and if they wanted to follow up on a data set or technique to find the matching worked examples in the project directory of our book support materials git repository.
Some readers want to work much closer to the sequence in the book. To make working along with book easier we extracted all book examples and shared them with our readers (in a Github directory, and a downloadable zip file, press “Raw” to download). The direct extraction from the book guarantees the files are in sync with our revised book. However there are trade-offs, sometimes (for legibility) the book mixed input and output without using R’s comment conventions. So you can’t always just paste everything. Also for a snippet to run you may need some libraries, data and results of previous snippets to be present in your R environment.
To help these readers we have added a new section to the book support materials: knitr markdown sheets that work all the book extracts from each chapter. Each chapter and appendix now has a matching markdown file that sets up the correct context to run each and every snippet extracted from the book. In principle you can now clone the entire zmPDSwR repository to your local machine and run all the from the CodeExamples directory by using the RStudio project in RunExamples. Correct execution also depens on having the right packages installed so we have also added a worksheet showing everything we expect to see installed in one place: InstallAll.Rmd (note some of the packages require external dependencies to work such as a C compiler, curl libraries, and a Java framework to run).
Our most recent article was a dynamic programming solution to the A/B test problem. Explicitly solving such dynamic programs gets long and tedious, so you are well served by finding and introducing clever invariants to track (something better than just raw win-rates). That clever idea is called “sequential analysis” and was introduced by Abraham Wald (somebody we have written about before). If you have ever heard of a test plan such as “first process to get more than 30 wins ahead of the other is the one we choose” you have seen methods derived from Wald’s sequential analysis technique.
“Data Science” is obviously a trendy term making it way through the hype cycle. Either nobody is good enough to be a data scientist (unicorns) or everybody is too good to be a data scientist (or the truth is somewhere in the middle).
And there is a quarter that grumbles that we are merely talking about statistics under a new name (see here and here).
It has always been the case that advances in data engineering (such as punch cards, or data centers) make analysis practical at new scales (though I still suspect Map/Reduce was a plot designed to trick engineers into being excited about ETL and report generation).
In this spirit next week we will write about the sequential analysis solution for A/B-testing, invented in the 1940s by one of the greats of statistics and operations research: Abraham Wald (whom we have written about before).
What stands out in these presentations is: the simple practice of a static test/train split is merely a convenience to cut down on operational complexity and difficulty of teaching. It is in no way optimal. That is, using slightly more complicated procedures can build better models on a given set of data.
Suggested static cal/train/test experiment design from vtreat data treatment library.