A US based startup wanted to develop a data management platform to process TV viewing records from different data sources and classify users into different segments based on their viewing pattern. This data could then be used for providing specific program recommendations and advertisements to the users.
The developed platform was expected to have the following features
- Provide recommendations for the programs and advertisement that a particular user might be interested in.
- Generate aggregation reports on the viewing of a universe of TV viewers
- Classify unknown users into different segments based on their viewing habits
Rare Mile worked with the client's team to collect and analyze TV viewing data available from different sources and use the same to create multiple reports for business processing. The solution also uses scalable machine learning algorithms to understand about the users based on their viewing patterns and provide recommendations.
Rare Mile proposed and implemented a solution with the following features:
- Use of Big Data technologies to process extremely large scale of data
- Applied machine learning techniques for classification, clustering and recommendations
- Batch and real time distributed processing for fast processing of data.
- Designed, implemented and successfully delivered the first two phases of the application on time.
- Provided optimally performing application to process large amount of data and high accuracy of the recommendations generated.
- Employed iterative development to absorb client's feedback regularly during the application development
- Used a research oriented approach for trying different candidate solution approaches before deciding the final solution
This project involved reading and analysing TV viewing information of millions and use that to provide program and advertisement recommendations.
- Client: US Based Startup
- Execution Time: 12 Weeks
- Category: Algorithms & Machine Learning