Measure TV viewership: Minimum Viable Product - GVC
Three tech students and GVC participants worked on a big global data project aiming at enhancing the evaluation of TV viewership. Based on the project, Hybrid’s clients would receive data about market overview directly from individual TV sets describing user behavior, effectivity, and commercial impact of TV ads.
We worked with more than 500.000 TV devices located in the Czech Republic, and three operating databases with raw data about the 1) playout data (broadcasters and shows), 2) users and devices identification related data (Hybrid ID data), and 3) users sessions data for the validation and creation of the minimum viable product.
Accomplished:
- Data architecture: 3 data models, 3 database connectors (ETL), and 5 data tables.
- BI processes: 6 case scenarios.
- Real scenarios to be executed: 7/5 scenarios finished.
- Data validation: 11/6 finished.
- Anomaly detection and forecasting with ML algorithms: forecast: 70%% done [2/3 tasks].
Highlights: