On a number of occasions, I've spoken to prospective clients and observed others making the following mistake. They've set up an A/B test or bandit system using relatively standard statistical methodology. Unfortunately, their sample size is prohibitively low. In order to "fix" this, they tweak the methodology until the …

Several years back, I wrote an article advocating in favor of using bandit algorithms. In retrospect, the article I wrote was incorrect, and I should have phrased it differently. I made no mathematical mistakes in the article. Every fact I said is true. But the implications of this article and …

Archiset - a wonderful set of architectural and movie illustrations. Actually look at nearly everything by Federico Babina.

Bayesianism and Causality, or Why I am only a Half-Bayesian. Great article introducing issues of causality in scientific inference. Most important point - "Causal assumptions, in contrast [to statistical assumptions], cannot be verified even …

Over the past year or so, there have been a number of articles discussing the use of one tailed vs two tailed A/B tests. For example, How Optimizely (Almost) Got Me Fired. The use of a one or two tailed test is an important issue. What every analysis I've …

It's becoming very popular to dislike Uber. Their executives are a bunch of jerks. They are eliminating the rents that certain politically powerful individuals were previously collecting. And moreover, occasionally Uber drivers do bad things, things that ordinary drivers didn't do many times in 2014 alone. This post is a …

So you've jumped onboard the A/B testing bandwagon. You've just run an A/B test comparing the site redesign to the old version. Unfortunately the redesign did not differ in a statistically significant way from the old version. At this point, a variety of conversion rate experts will tell …

It was a stupid advertisement like the one you see above that made me realize: I need to be on HTTPS. Seeing Amitabh Bachchan hawking life insurance or soap is an everyday occurrence. But what was surprising about this patricular ad was that it appeared on my blog. This is …

An Homage to London Brutalism

Multiple comparisons don't have to feel like fishing

How Naive AB Testing Goes Wrong and How to Fix It. Nice to see that someone is using a proper decision rule. Note: my conflict of interest is obvious to anyone who reads the articles.

Timeseries: Cointegration …

The two sides of my programming world have long suffered a disconnect. When I build a robust low latency system, I typically write it in a language like Scala. I make great use of type safety to reduce the number of errors I've made. On the other hand, when I …

An illustrated cross section of Kowloon Walled City

Probable points and credible intervals. A great article about things to do with a posterior once you compute it.

How misaligning data can increase performance 12X by reducing cache misses

Scalazy streams - assorted tricks to make your Scala 2.11.x collections …