Question: VOD specific data: genre, producer, play time, number of views The company wants to initiate an effective recommendation service to its subscribers. The purpose of the service is to increase the satisfaction of current subscribers with the company. As the project leader, you are going to develop a recommendation model for the service. Note that your model is concerning WHAT and WHEN to recommend.
TV program specific data: genre, producer, viewer rate (popularity) over time
User-program specific data: cable TV programs recorded in the set-top box, watched VOD
User-day specific data: time spent watching VOD (e.g., from 2pm to 5pm on Mar 18, from 1am to 3am and from 1pm to 3pm on Mar 19)
User-specific data: gender, age, date of subscription, number of family members in the house hold, number of subscribing TVs
Assume that you have joined Hong Kong Broadband after graduation. The company provides cable TV and video-on-demand (VOD)1 services to its subscribers. The company collects a variety of data about its subscribers and products such as followings:
1) Which variables will you use for the recommendation model? Why do you think those variables are useful to build up the recommendation model?
2) How will you evaluate your recommendation model? Describe your strategy.