Robert Chang piece on how data products have always been an instrumental part of Airbnb’s service. However, engineers have long recognized that it’s costly to make data products.
Robert describes how auto machine learning tools worked together to expedite the modeling process and hence lower the overall development costs for a specific use case of life time value modeling — predicting the value of homes on Airbnb.
He describes:
- What customer life time value is (LTV)
- Machine learning workflow or LTV modeling
- Feature engineering they use (Zipline)
- Prototyping and training in Python
- How they perform model selection
and more. Example of code are included. Great article!
[Read More]