Zhang W, Chen J (2019) Digging for Gold with Data Analytics at Chow Tai Fook

The Chow Tai Fook (CTF) case introduces an important challenge for data analytics in the jewelry or luxury product industry. It is difficult to provide an accurate demand prediction for a product at CTF. The reasons are mainly three-fold. First, the jewelry products are of high value and slow-moving, with mostly single digit sales numbers over a year. Second, the demand of a product partly depends on the availability of other products, due to customers’ demand substitution behavior. Lastly, there are 16 product categories with more than 15,000 distinct products offered by CTF, but each customer only chooses from a small subset of these products. In this case, traditional statistical methods such as linear regression or time series forecasting are not useful. I teach students how to use discrete choice models to tackle this challenge.