Scribd
Upload
1K views2 pages

4-6 Analytics Lifecycle Case Study Netflix

Netflix uses a personalized recommendation system to connect users to movies they love. Originally called Cinematch, it predicts if users will enjoy movies based on past likes and dislikes. To improve recommendations for its new online streaming service, Netflix held a competition won by BellKor to transform Cinematch. The updated system considers data like movie genres, reviews, and user geography to provide tailored suggestions. Throughout the years, Netflix's analytics team has developed algorithms using data like watch histories and user demographics to continuously refine recommendations and enrich the streaming experience.

Uploaded by

Luis Alfonso Dañez
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
1K views2 pages

4-6 Analytics Lifecycle Case Study Netflix

Netflix uses a personalized recommendation system to connect users to movies they love. Originally called Cinematch, it predicts if users will enjoy movies based on past likes and dislikes. To improve recommendations for its new online streaming service, Netflix held a competition won by BellKor to transform Cinematch. The updated system considers data like movie genres, reviews, and user geography to provide tailored suggestions. Throughout the years, Netflix's analytics team has developed algorithms using data like watch histories and user demographics to continuously refine recommendations and enrich the streaming experience.

Uploaded by

Luis Alfonso Dañez
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

1.

Discovery
Netflix is all about connecting its users or customers to the movies or films they love
through a personalized recommendation system. When the company was still a mail-based
DVD rental service company, this recommendation system was known as Cinematch. It’s
designed to predict whether the customers will enjoy the movie based on their previous
liked or disliked movies. The problem is that the company had to rely on third party postal
services to deliver its DVDs to its customers. This resulted to heavy losses, so they soon
shifted to online streaming service. For its new service to be successful, the Netflix team
held a competition that will help improve their existing recommendation system,
Cinematch. With the win from BellKor team, Netflix was able to transform its existing model
to respond with its new parameters and objectives as an online streaming service company.
The team was able to identify the key stakeholders that will benefit from its new model. This
includes the people who will use their online streaming service to find the movies they want
and be able to find other movies they might like based on their preferences or interests. The
team considered various data sources in developing its recommendation system such as
movie genre, duration, and reviews.

2. Data Preparation
The Netflix’s early recommendation system, the Cinematch, is a good quality source of
data. It is one of the best data sources the team can explore in developing a new
recommendation system fit for their new online streaming service. The team can use and
learn from the old model’s strength and weaknesses. As Netflix continues to expand its
reach to more users, the team constantly assessed the data quality it needs for them to be
able to deliver and recommend movies that suits the preferences and interests of its unique
users. The team realized that they also had to consider the user’s geography, race or
nationality in developing their new recommendation system.

3. Model Planning
After preparing and identifying the good quality data and possible data sources, the
team moves towards the planning stage of their new recommendation system. With the
many algorithms the Netflix company has invested for back in the 2006 competition, the
BellKor team’s algorithm stands out as it was able to provide suggestions based on what the
users’ needs or preferences.
4. Model Building
Throughout the years, the analytics team came up with different models of algorithms
that helped them with their data processing, with algorithms such as K-nearest algorithm
and various ranking algorithms, but they were all focused on the dependency of subscribers,
specifically factors that impact the viewer’s enjoyment. To predict the best algorithm
effective for their company, the team made use of A/B testing and interleaving techniques
to predict if new and proposed algorithms are better than their current one. Furthermore, it
helped them predict which algorithm will be most recommended for the consumers. To
overcome this, the Netflix company stored and kept batches of data, including the
subscribers’ watch histories, the time of day a certain movie was watched, and a user’s age
and gender. These are data points that helped them to improve with their current service.
5. Communicate Results
With reports of their company continuously gathering more users and subscribers during
the years, their findings lead to their conclusion that Netflix is reaching at a global level already.
From this finding, their team concluded that they had to diversify their services, and so they
have also recorded their transactions with their subscribers, including their geography and
language as structuralized data. Using data-driven processes, the team tried to record other
data such as star-rating systems and review forms so that they will be aware of the
recommendations made by the consumers to see if they still need to improve with the service
they are providing.
6. Operationalize
Up until the present time, the analytics team of Netflix still implements data-driven
experiments and algorithms in developing newer and more effective models to provide users
with enriched content while they stream and watch their chosen movies. With their current
process, what the analytics team generally does is a cycle of collecting data from their
subscribers by the movies that they watch, and use that particular data to sort out movie
categories and genres such as drama, sci-fi, action, and comedies. Furthermore, they created
recommendation systems that continuously provide suggestions for movies and shows for
viewers that are related to the current movie that they are watching. These databases that they
gather will be used for further insights. Eventually, they would either use this database to detect
bugs and errors in their systems or to formulate new data that can be used to further improve
their services towards their subscribers. Lastly, predicting future data will be done by the team
to determine whether their current algorithm will still be relevant in long-term or short-term
situations.

You might also like