from 13:30 to 14:405
Many phenomena that extend along time require a specific approach to determine the effects over time of a given set of events, for instance online advertising campaigns.
Causal effects in observational time series can be identified by estimating a synthetic control.
This can be done using structural time-series models and suitable predictors.
The CausalImpact R package implements one approach to this.
GoogleSenior Quantitative Analyst