Notes on Data science in media industry
After about 16-year working in Institute for Infocomm Research, A*STAR, Singapore https://www.a-star.edu.sg/i2r as a research scientist, I realized I need make a change in my life, to learn how industry exploits machine learning and pattern recognition technology to build product and solve business problem. In 2019, I have a chance to join a media company to lead a data science team. In the following posts, I will summarize my learning journey of building data product & machine learning models to solve business problems. The main topics includes:
- User profile (data product)
- User personal information related prediction
- Gender prediction
- Age prediction
- Race prediction
- Media content (news reading, video view) related prediction
- Traffic prediction
- How to know content popularity in advance content publishing for efficient resource planning
- Auto content tag: label news content using IAB (https://www.iab.com/videos/iab-there/) tag set, a business related semantic tag.
- Traffic prediction
- Digital optimizer
- Personalized advertisement targeting
- How to improve advertisement performance (CTR, click through rate) based on third-party data collection toolkit (e.g. Adobe cookie, Google ads performance data)
Data science is to exploit machine learning and pattern recognition technology to solve business problems in industry. Every company is different in terms of industry, operation environment, data available, and business problems. Data science is data-driven approach to solve business problem, thus the most important stages are 1) to transform business problem to machine learning problem (discuss with business owner to understand their requirement), and 2) to collect correct data to build ML model.