
Time series decomposition for incrementality and forecasting with A/B testing
Last week I had the delightful honor of being the last ever speaker at eMetrics London. Don’t worry – it’ll be back, rebranded as Marketing Analytics World: https://twitter.com/jimsterne/status/918798356343861248
I talked about how we’ve used time series decomposition (via R) to demonstrate how to quantify the longer term effects of A/B tests (or any test) with reasonable levels of confidence.
Here’s the deck along with the game show intro 🙂