Serendipity

A personalization, promotion and reccomendation algorithm that helps personalize Rogers TV Services by leveraging exponential technologies like machine learning and natural language processing.

Comment

Inspiration

Low user engagement of Rogers TV Main drivers:

  • Variety + Flexibility (81% of respondents stated variety makes a good TV service provider)
  • Pricing (60% of respondents stated low prices make a good provide)

After collecting some data from surveys, we found that:

  • Netflix is popular
  • Most users are price sensitive
  • Older people willing to pay more
  • Older viewers watch least often

From looking at reviews for the Rogers Anyplace TV app we discovered that... Slow Errors Crashing Overall poorly designed

What it does

Consumers create customized bundles based on channels of interest Flexible pricing based on channels selected Suggestions prompted via API / ML Layer Long-Term: Streaming across internet devices

Recommender: Make recommendations to personalize the user experience

How we built it

We collect user data and used the rogers data. Then, we cleaned the data. From there, we used MonkeyLearn's NLP API to use a variety of fo classifiers and extractors to detect sentiment and gain insights. In addition, we worked with the Sports Radar API to get some contextual information in a variety of sports and teams to give personalized promotions/bundles

Challenges we ran into

The rogers data set was very limited and not cleaned. The lack of data was the biggest problem in getting any sort of personalization.

Accomplishments that we're proud of

Great team effort and working with a variety of APIs. We all learned something new!

Share this project:

Updates