This includes four main steps
- Understanding the data, i.e. trends, seasonality etc
- Detecting the changepoint of the demand pattern, i.e. at what period did this sudden shock start affecting the business.
- Capturing the causal impact of this sudden shock on the business
- Adding this impact as an index to the machine learning forecasted demands.
All steps are captured in this notebook. You can also find the article associated to this project here for more explaination.