Netflix's global subscriber base has reached an impressive milestone, surpassing *** million paid subscribers worldwide in the fourth quarter of 2024. This marks a significant increase of nearly ** million subscribers compared to the previous quarter, solidifying Netflix's position as a dominant force in the streaming industry. Adapting to customer losses Netflix's growth has not always been consistent. During the first half of 2022, the streaming giant lost over *** million customers. In response to these losses, Netflix introduced an ad-supported tier in November of that same year. This strategic move has paid off, with the lower-cost plan attracting ** million monthly active users globally by November 2024, demonstrating Netflix's ability to adapt to changing market conditions and consumer preferences. Global expansion Netflix continues to focus on international markets, with a forecast suggesting that the Asia Pacific region is expected to see the most substantial growth in the upcoming years, potentially reaching around **** million subscribers by 2029. To correspond to the needs of the non-American target group, the company has heavily invested in international content in recent years, with Korean, Spanish, and Japanese being the most watched non-English content languages on the platform.

Context

Netflix, Inc. is an American media company engaged in paid streaming and the production of films and series.

Market cap

Market capitalization of Netflix (NFLX)

Market cap: $517.08 Billion USD

As of June 2025 Netflix has a market cap of $517.08 Billion USD. This makes Netflix the world's 19th most valuable company by market cap according to our data. The market capitalization, commonly called market cap, is the total market value of a publicly traded company's outstanding shares and is commonly used to measure how much a company is worth.

Revenue

Revenue for Netflix (NFLX)

Revenue in 2025: $40.17 Billion USD

According to Netflix's latest financial reports the company's current revenue (TTM ) is $40.17 Billion USD. In 2024 the company made a revenue of $39.00 Billion USD an increase over the revenue in the year 2023 that were of $33.72 Billion USD. The revenue is the total amount of income that a company generates by the sale of goods or services. Unlike with the earnings no expenses are subtracted.

Earnings

Earnings for Netflix (NFLX)

Earnings in 2025 (TTM): $11.31 Billion USD

According to Netflix's latest financial reports the company's current earnings are $40.17 Billion USD. In 2024 the company made an earning of $10.70 Billion USD, an increase over its 2023 earnings that were of $7.02 Billion USD. The earnings displayed on this page is the company's Pretax Income.

End of Day market cap according to different sources

On Jun 12th, 2025 the market cap of Netflix was reported to be:

$517.08 Billion USD by Yahoo Finance

$517.08 Billion USD by CompaniesMarketCap

$517.21 Billion USD by Nasdaq

Content

Geography: USA

Time period: May 2002- June 2025

Unit of analysis: Netflix Stock Data 2025

Variables

Acknowledgements

This dataset belongs to me. I’m sharing it here for free. You may do with it as you wish.

In this post, you'll see how the Netflix platform is evolving, how many users Netflix has and how they perform against the growing competition.

1) Data Introduction • The Netflix Movies and TV Shows Dataset contains various metadata on movies and TV shows available on Netflix. • Key features include the title, director, cast, country, date added, release year, rating, genre, and total duration (in minutes or number of seasons) of the content.

2) Data Utilization (1) Characteristics of the Netflix Movies and TV Shows Dataset • This dataset helps in understanding content trends and markets, as well as analyzing global preferences and changing consumer tastes. • It is useful for analyzing the characteristics of content available in different countries, including genre, cast, director, and more.

(2) Applications of the Netflix Movies and TV Shows Dataset • Content Analysis: Analyze how Netflix's content is distributed, and understand preferences based on genre or country. • Recommendation System Development: Develop algorithms that recommend similar content based on user viewing patterns. • Market Analysis: Identify which content is popular in different countries and analyze if Netflix focuses more on specific countries or genres.

I extracted this data to find the unpopular movies on Netflix. The dataset I used here comes directly from Netflix movies data, which consists of 4 text data files, each file contains over 20M rows, over 4K movies, and 400K, customers. Altogether over are 17K movies and 500K+ customers!

I made some modifications and I extracted the e df_avgRating_with_usersCount.csv from the original data after applying some mathematical operations to get the average ratings and the count of users who made the ratings for each movie in movie_id below. Feel free to browse and use the data within your notebooks.

Here you could find my previous notebook on Kaggle to extract the dataset

Analysis of ‘1000 Netflix Shows’ provided by Analyst-2 (analyst-2.ai), based on source dataset retrieved from https://www.kaggle.com/chasewillden/netflix-shows on 28 January 2022.

--- Dataset description provided by original source is as follows ---

Context

Netflix in the past 5-10 years has captured a large populate of viewers. With more viewers, there most likely an increase of show variety. However, do people understand the distribution of ratings on Netflix shows?

Content

Because of the vast amount of time it would take to gather 1,000 shows one by one, the gathering method took advantage of the Netflix’s suggestion engine. The suggestion engine recommends shows similar to the selected show. As part of this data set, I took 4 videos from 4 ratings (totaling 16 unique shows), then pulled 53 suggested shows per video. The ratings include: G, PG, TV-14, TV-MA. I chose not to pull from every rating (e.g. TV-G, TV-Y, etc.).

Acknowledgements

The data set and the research article can be found at The Concept Center

Inspiration

I was watching Netflix with my wife and we asked ourselves, why are there so many R and TV-MA rating shows?

--- Original source retains full ownership of the source dataset ---

Analysis of ‘Netflix Shows’ provided by Analyst-2 (analyst-2.ai), based on source dataset retrieved from https://www.kaggle.com/yamqwe/netflix-showse on 13 February 2022.

--- Dataset description provided by original source is as follows ---

About this dataset

Background

Netflix in the past 5-10 years has captured a large populate of viewers. With more viewers, there most likely an increase of show variety. However, do people understand the distribution of ratings on Netflix shows?

Netflix Suggestion Engine

Because of the vast amount of time it would take to gather 1,000 shows one by one, the gathering method took advantage of the Netflix’s suggestion engine. The suggestion engine recommends shows similar to the selected show. As part of this data set, I took 4 videos from 4 ratings (totaling 16 unique shows), then pulled 53 suggested shows per video. The ratings include: G, PG, TV-14, TV-MA. I chose not to pull from every rating (e.g. TV-G, TV-Y, etc.).

Source

Access to the study can be found at The Concept Center

This dataset was created by Chase Willden and contains around 1000 samples along with User Rating Score, Rating Description, technical information and other features such as: - Release Year - Title - and more.

How to use this dataset

• Analyze User Rating Size in relation to Rating
• Study the influence of Rating Level on User Rating Score
• More datasets

Acknowledgements

If you use this dataset in your research, please credit Chase Willden

Start A New Notebook!

--- Original source retains full ownership of the source dataset ---

Context

Netflix held the Netflix Prize open competition for the best algorithm to predict user ratings for films. The grand prize was $1,000,000 and was won by BellKor's Pragmatic Chaos team. This is the dataset that was used in that competition.

Content

This comes directly from the README:

TRAINING DATASET FILE DESCRIPTION

The file "training_set.tar" is a tar of a directory containing 17770 files, one per movie. The first line of each file contains the movie id followed by a colon. Each subsequent line in the file corresponds to a rating from a customer and its date in the following format:

CustomerID,Rating,Date

• MovieIDs range from 1 to 17770 sequentially.
• CustomerIDs range from 1 to 2649429, with gaps. There are 480189 users.
• Ratings are on a five star (integral) scale from 1 to 5.
• Dates have the format YYYY-MM-DD.

MOVIES FILE DESCRIPTION

Movie information in "movie_titles.txt" is in the following format:

MovieID,YearOfRelease,Title

• MovieID do not correspond to actual Netflix movie ids or IMDB movie ids.
• YearOfRelease can range from 1890 to 2005 and may correspond to the release of corresponding DVD, not necessarily its theaterical release.
• Title is the Netflix movie title and may not correspond to titles used on other sites. Titles are in English.

QUALIFYING AND PREDICTION DATASET FILE DESCRIPTION

The qualifying dataset for the Netflix Prize is contained in the text file "qualifying.txt". It consists of lines indicating a movie id, followed by a colon, and then customer ids and rating dates, one per line for that movie id. The movie and customer ids are contained in the training set. Of course the ratings are withheld. There are no empty lines in the file.

MovieID1:

CustomerID11,Date11

CustomerID12,Date12

...

MovieID2:

CustomerID21,Date21

CustomerID22,Date22

For the Netflix Prize, your program must predict the all ratings the customers gave the movies in the qualifying dataset based on the information in the training dataset.

The format of your submitted prediction file follows the movie and customer id, date order of the qualifying dataset. However, your predicted rating takes the place of the corresponding customer id (and date), one per line.

For example, if the qualifying dataset looked like:

111:

3245,2005-12-19

5666,2005-12-23

6789,2005-03-14

225:

1234,2005-05-26

3456,2005-11-07

then a prediction file should look something like:

111:

3.0

3.4

4.0

225:

1.0

2.0

which predicts that customer 3245 would have rated movie 111 3.0 stars on the 19th of Decemeber, 2005, that customer 5666 would have rated it slightly higher at 3.4 stars on the 23rd of Decemeber, 2005, etc.

You must make predictions for all customers for all movies in the qualifying dataset.

THE PROBE DATASET FILE DESCRIPTION

To allow you to test your system before you submit a prediction set based on the qualifying dataset, we have provided a probe dataset in the file "probe.txt". This text file contains lines indicating a movie id, followed by a colon, and then customer ids, one per line for that movie id.

MovieID1:

CustomerID11

CustomerID12

...

MovieID2:

CustomerID21

CustomerID22

Like the qualifying dataset, the movie and customer id pairs are contained in the training set. However, unlike the qualifying dataset, the ratings (and dates) for each pair are contained in the training dataset.

If you wish, you may calculate the RMSE of your predictions against those ratings and compare your RMSE against the Cinematch RMSE on the same data. See http://www.netflixprize.com/faq#probe for that value.

Acknowledgements

The training data came in 17,000+ files. In the interest of keeping files together and file sizes as low as possible, I combined them into four text files: combined_data_(1,2,3,4).txt

The contest was originally hosted at http://netflixprize.com/index.html

The dataset was downloaded from https://archive.org/download/nf_prize_dataset.tar

Inspiration

This is a fun dataset to work with. You can read about the winning algorithm by BellKor's Pragmatic Chaos here

Context

Netflix in the past 5-10 years has captured a large populate of viewers. With more viewers, there most likely an increase of show variety. However, do people understand the distribution of ratings on Netflix shows?

Content

Because of the vast amount of time it would take to gather 1,000 shows one by one, the gathering method took advantage of the Netflix’s suggestion engine. The suggestion engine recommends shows similar to the selected show. As part of this data set, I took 4 videos from 4 ratings (totaling 16 unique shows), then pulled 53 suggested shows per video. The ratings include: G, PG, TV-14, TV-MA. I chose not to pull from every rating (e.g. TV-G, TV-Y, etc.).

Acknowledgements

The data set and the research article can be found at The Concept Center

Inspiration

I was watching Netflix with my wife and we asked ourselves, why are there so many R and TV-MA rating shows?

This dataset addresses the common issue of finding quality content amidst a vast catalogue, specifically on Netflix. It aims to help users discover underrated content and hidden gems. The dataset aggregates information from multiple sources, including Netflix itself, Rotten Tomatoes, and IMDb, combining various attributes to provide deeper insights into content quality and characteristics. A unique "Hidden Gem Score" is included, calculated based on low review counts and high user ratings, making it easier to identify valuable content that might otherwise be overlooked. This dataset powers the FlixGem.com platform, a related project designed for interactive exploration.

Columns

The dataset includes several key columns to facilitate detailed analysis of Netflix content: * Title: The name of the movie or series. * Genre: Hundreds of genre classifications for the content. * Tags: Thousands of detailed tags describing the content. * Languages: Languages available for the content, including English and many others. * Series or Movie: Indicates whether the content is a TV series or a movie. * Hidden Gem Score: A calculated metric based on low review counts and high ratings to identify hidden gems. * Country Availability: Information on Netflix country availability for the content. * Runtime: The duration of the series or movie. * Director: The director of the content. * Writer: The writer of the content.

Distribution

The data files are typically in CSV format. This dataset is regularly updated, with monthly revisions to ensure freshness. It was last updated in early April 2021. The dataset is version 1.0. While specific total row or record counts are not provided, some columns feature a considerable number of unique values, such as over 15,000 unique genres and over 13,000 unique languages.

Usage

This dataset is ideal for various analytical and exploratory applications, including: * Finding correlations between ratings, actors, directors, and box office performance. * Identifying patterns related to content quality based on characteristics like language and genre. * Discovering hidden gems across different regions. * Interactive browsing and knowledge discovery through platforms like FlixGem.com, which is powered by this very dataset. * Developing machine learning models for content recommendation or classification.

Coverage

The dataset offers global regional coverage, with a specific column indicating Netflix country availability for content. It focuses on recent Netflix data, with monthly updates provided. The last update was in early April 2021. The content spans a wide range of genres and includes various languages, with English being a significant portion. Runtime varies, with a large percentage of content being 1-2 hours long, followed by content under 30 minutes.

License

CCO

Who Can Use It

This dataset is designed for anyone interested in delving deeply into Netflix content, including: * Data analysts looking to unearth trends and insights. * Researchers studying media consumption patterns or content quality. * Developers creating recommendation engines or content discovery tools. * Machine learning practitioners building models for classification or prediction. * Content strategists seeking to understand what makes content resonate. * Individuals simply curious about finding their next favourite show or movie.

Dataset Name Suggestions

• Netflix Hidden Gems Dataset
• Netflix Content Quality & Discovery Data
• Global Netflix Catalogue Insights
• Curated Netflix Content Attributes
• Netflix Underrated Content Analysis

Attributes

Original Data Source: Latest Netflix data with 26+ joined attributes

This is a reduced dataset from a much larger Netflix's movie ratings database, for use in collaborative filtering, recommendation systems, and related applications.

Any particular user has rated only a fraction of the movies, so the data matrix is only partially filled. The goal here is to fill all the remaining entries of the matrix, and then compare with the complete test matrix.

This dataset provides a detailed list and metadata for approximately 7,000 TV shows and movies available on Netflix as of June 2021. Sourced from the IMDB website, it offers insights into content characteristics, popularity, and categorisation, making it suitable for various analytical and machine learning applications.

Columns

• imdb_id: A unique identifier for each show or movie.
• title: The title of the television programme or film.
• popular_rank: The ranking assigned by IMDB based on popularity.
• certificate: Age certifications received by the content; it is noted that many values may be null.
• startYear: The year the show was first broadcast or the film was released.
• endYear: The year a show concluded, if applicable.
• episodes: The total number of episodes in a series; for films, this value is 1.
• runtime: The running time of the content.
• type: Specifies whether the content is a 'Movie' or 'Series'.
• orign_country: The country of origin for the show or movie.
• language: The primary language of the content.
• plot: A synopsis of the show or movie.
• summary: A concise summary of the story.
• rating: The average user rating for the content.
• numVotes: The total number of votes received for the content's rating.
• genres: The genre(s) to which the show or movie belongs.
• isAdult: A binary indicator (1 for adult content, 0 otherwise).
• cast: The main cast members listed in a suitable format.
• image_url: A link to the poster image for the content.

Distribution

The dataset is typically provided as a CSV file, specifically named netflix_list.csv. It contains approximately 7,000 records, with 7,008 unique identifiers for shows and movies. This dataset is listed as version 1.0 and was added to the platform on 11 June 2025.

Usage

This dataset is ideally suited for developing recommender systems, performing natural language processing (NLP) tasks on plot summaries, and conducting market analysis of entertainment content. It can be used to explore trends in movie and TV show production, analyse viewer preferences, and facilitate content categorisation efforts.

Coverage

The dataset offers global coverage, with information on content originating from various countries. The startYear of content spans from 1932 to 2022, with the majority of content released between 2004 and 2022. The endYear ranges from 1969 to 2022, with most data concentrated from 2011 to 2022. It includes age certification information and an indicator for adult content, allowing for demographic considerations related to content suitability.

License

CCO

Who Can Use It

This dataset is valuable for data scientists and machine learning engineers working on content recommendation engines or text analysis projects. It is also beneficial for researchers studying media consumption patterns and entertainment industry analysts interested in exploring the Netflix content catalogue programmatically.

Dataset Name Suggestions

• Netflix Content Metadata (June 2021)
• Global Netflix Catalogue
• Netflix IMDB Dataset
• Streaming Content Insights (Netflix)
• Netflix Movie and TV Show Archive

Attributes

Original Data Source:Netflix Movie and TV Shows (June 2021)

This dataset offers a comprehensive historical record of Netflix’s stock price movements, capturing the company’s financial journey from its early days to its position as a global streaming giant.

From its IPO in May 2002, Netflix (Ticker: NFLX) has transformed from a DVD rental service to a powerhouse in on-demand digital content. With its disruptive innovation, strategic shifts, and global expansion, Netflix has seen dramatic shifts in stock prices, reflecting not just market trends but also cultural impact. This dataset provides a window into that evolution.

What’s Included?

Each row in this dataset represents daily trading activity on the stock market and includes the following columns:

• Date – The trading day (from 2002 onward)
• Open – Stock price when the market opened
• High – Highest trading price of the day
• Low – Lowest trading price of the day
• Close – Final price at market close
• Adj Close – Closing price adjusted for splits and dividends
• Volume – Number of shares traded that day

The data is structured in CSV format and is clean, easy to use, and ready for immediate analysis.

Why Use This Dataset?

Whether you're learning data science, building a financial model, or exploring machine learning in the real world, this dataset is a goldmine of insights. Netflix's market history includes:

• Periods of explosive growth during digital transformation
• Volatility during market crashes and global events (e.g., 2008, COVID-19)
• Strategic pivots such as the shift to original content
• Market reactions to earnings, acquisitions, and subscriber milestones

This makes the dataset ideal for:

• Time-series forecasting (ARIMA, Prophet, LSTM)
• Technical and trend analysis (moving averages, RSI, Bollinger Bands)
• Predictive modeling with machine learning
• Investment simulation projects
• Stock market visualization and storytelling
• Financial dashboards (Tableau, Power BI, Streamlit, etc.)

Who Can Use It?

This dataset is designed for:

• Aspiring data scientists practicing EDA and modeling
• Financial analysts and traders exploring trends
• AI researchers working on time-series models
• Students building ML projects
• Developers creating stock visualization tools
• Kaggle competitors seeking real-world datasets

Data Source & Credits

The dataset is derived from publicly available historical stock price data, such as Yahoo Finance, and has been cleaned and organized for educational and research purposes. It is continuously maintained to ensure accuracy.

Start Exploring

Netflix’s rise is more than just a business story — it’s a data-driven journey. With this dataset, you can analyze the company’s stock behavior, train models to predict future trends, or simply visualize how tech reshapes the market.

The Netflix Prize was a competition devised by Netflix to improve the accuracy of its recommendation system. To facilitate this Netflix released real ratings about movies from the users (voters) of the system. Any set of movies can be transformed into an election via a process outlined by Mattei, Forshee, and Goldsmith.This data set includes all 5 candidate elections with at least 350 voters generated by this process from 300 randomly chosen movies. Extending beyond prior work by Mattei et al. we allow for weak preferences, i.e., a voter is indifferent between a set of movies if he assigns the same rating to each of them. Thus, there are 541 possibilities to rank a given set of five movies.The archive is gzip compressed and includes 165,672 elections in PrefLib.org's TOC file format (Orders with Ties - Complete List).

Overview

Streaming Services Statistics: Streaming services have transformed the entertainment landscape, revolutionizing how people consume content.

The advent of high-speed internet and the proliferation of smart devices have fueled the growth of these platforms, offering a wide array of movies, TV shows, music, and more, at the viewers' convenience.

This introduction provides an overview of key statistics that shed light on the impact, trends, and challenges within the streaming industry.

Context

The art of forecasting stock prices has been a difficult task for many of the researchers and analysts. In fact, investors are highly interested in the research area of stock price prediction. For a good and successful investment, many investors are keen on knowing the future situation of the stock market. Good and effective prediction systems for the stock market help traders, investors, and analyst by providing supportive information like the future direction of the stock market. In this work, we present a recurrent neural network (RNN) and Long Short-Term Memory (LSTM) approach to predict stock market indices.

Content

Data from yahoo.com/finance Netflix historical price 12/16/2015 ~ 12/16/2019 daily price and volume. There are 7 columns; Date, open, high, low, close, volume, adj close (2001, 7) each of stock

Acknowledgements

I want to find relationship between volume and price.

Inspiration

Here is couple of things one could try out with this data:

One day ahead prediction: Rolling Linear Regression, ARIMA, Neural Networks, LSTM
Momentum/Mean-Reversion Strategies
Security clustering, portfolio construction/hedging

Your data will be in front of the world's largest data science community. What questions do you want to see answered?

DESCRIPTION

The dataset provided contains movie reviews given by Amazon customers. Reviews were given between May 1996 and July 2014.

Data Dictionary UserID – 4848 customers who provided a rating for each movie Movie 1 to Movie 206 – 206 movies for which ratings are provided by 4848 distinct users

Data Considerations - All the users have not watched all the movies and therefore, all movies are not rated. These missing values are represented by NA. - Ratings are on a scale of -1 to 10 where -1 is the least rating and 10 is the best.

Analysis Task - Exploratory Data Analysis:

Which movies have maximum views/ratings? What is the average rating for each movie? Define the top 5 movies with the maximum ratings. Define the top 5 movies with the least audience. - Recommendation Model: Some of the movies hadn’t been watched and therefore, are not rated by the users. Netflix would like to take this as an opportunity and build a machine learning recommendation algorithm which provides the ratings for each of the users.

Divide the data into training and test data Build a recommendation model on training data Make predictions on the test data

Industry data revealed that Slovakia had the most extensive Netflix media library worldwide as of July 2024, with over 8,500 titles available on the platform. Interestingly, the top 10 ranking was spearheaded by European countries. Where do you get the most bang for your Netflix buck? In February 2024, Liechtenstein and Switzerland were the countries with the most expensive Netflix subscription rates. Viewers had to pay around 21.19 U.S. dollars per month for a standard subscription. Subscribers in these countries could choose from between around 6,500 and 6,900 titles. On the other end of the spectrum, Pakistan, Egypt, and Nigeria are some of the countries with the cheapest Netflix subscription costs at around 2.90 to 4.65 U.S. dollars per month. Popular content on Netflix While viewing preferences can differ across countries and regions, some titles have proven particularly popular with international audiences. As of mid-2024, "Red Notice" and "Don't Look Up" were the most popular English-language movies on Netflix, with over 230 million views in its first 91 days available on the platform. Meanwhile, "Troll" ranks first among the top non-English language Netflix movies of all time. The monster film has amassed 103 million views on Netflix, making it the most successful Norwegian-language film on the platform to date.

In the fourth quarter of 2024, Netflix generated total revenue of over **** billion U.S. dollars, up from about *** billion dollars in the corresponding quarter of 2023. The company's annual revenue in 2024 amounted to around ** billion U.S. dollars, continuing the impressive year-on-year growth Netflix has enjoyed over the last decade. Netflix’s global position Netflix’s revenue has been heavily impacted by its ever-growing global subscriber base. The leading Netflix market is Europe, Middle East, and Africa, surpassing the U.S. and Canada in terms of subscriber count. Netflix has also significantly increased its licensed and produced content assets since 2016. Despite concerns among investors that the company’s content spend was negatively affecting cash flow, Netflix’s plans to amortize its content assets long-term along with generating revenue from other sources such as licensing and merchandise should ensure the company’s future profitability. Netflix’s original content Netflix is also fortunate in that many of its original shows have been a hit with consumers across the globe. Shows such as “Orange is the New Black,” “Black Mirror,” and “House of Cards” won the hearts of subscribers long ago, but newer content such as English-language shows “Bridgerton,” “Wednesday,” and “Stranger Things,” as well as local TV shows such as “Squid Game” have also been favorably reviewed and proved popular among users.

Context

Network monitoring and analysis of consumption behavior represents an important aspect for network operators allowing to obtain vital information about consumption trends in order to offer new data plans aimed at specific users and obtain an adequate perspective of the network. Over-the-top (OTT) media and communications services and applications are shifting the Internet consumption by increasing the traffic generation over the different available networks. OTT refers to applications that deliver audio, video, and other media over the Internet by leveraging the infrastructure deployed by network operators but without their involvement in the control or distribution of the content and are known by their large consumption of network resources.

Content

This dataset contains 1581 instances and 131 attributes on a single file. Each instance represents a user’s consumption profile which holds summarized information about the consumption behavior of the user related to the 29 OTT applications identified in the different IP flows captured in order to create the dataset

The OTT applications that the users interacted with during the capture experiment and were stored on the dataset are: Amazon, Apple store, Apple Icloud, Apple Itunes, Deezer, Dropbox, EasyTaxi, Ebay, Facebook, Gmail, Google suite, Google Maps, Browsing (HTTP, HTTP_Connect, HTTP_Download, HTTP_Proxy), Instagram, LastFM, Microsoft One Drive (MS_One_Drive), Facebook Messenger (MSN), Netflix, Skype, Spotify, Teamspeak, Teamviewer, Twitch, Twitter, Waze, Whatsapp, Wikipedia, Yahoo and Youtube.

Each application has 4 different types of attributes (quantity of generated flows, mean duration of the flows, average size of the packets exchanged on the flows and the mean bytes per second on the flows). These attributes summarizes the interaction that the user had with the respective OTT application in terms of consumption. Furthermore, the dataset contains the user’s IP address in network and decimal format which are used as user identifiers. Finally the User Group attribute represents the objective class (high consumption, medium consumption and low consumption) in which a user is classified considering his/her OTT consumption behavior. All of this information gives a total of 131 attributes.

For further information you can read and please cite the following papers:

Research Gate: https://www.researchgate.net/publication/326150046_Personalized_Service_Degradation_Policies_on_OTT_Applications_Based_on_the_Consumption_Behavior_of_Users

Springer: https://link.springer.com/chapter/10.1007/978-3-319-95168-3_37

Research Gate: https://www.researchgate.net/publication/335954240_Consumption_Behavior_Analysis_of_Over_The_Top_Services_Incremental_Learning_or_Traditional_Methods

IEEExplore: https://ieeexplore.ieee.org/document/8845576

Attribute Description

The structure of the attributes and its definition is presented below:

• Source.Decimal: This attribute holds the user’s IP address in decimal format and it is mainly used as a user identifier.

• Source.IP: This attribute holds the user’s IP address in network format (e.g., 192.168.14.35) and as in the previous case its main function is to work as a user identifier.

• Application-Name.Flows: This type of attributes hold the information about the quantity of IP flows that a user generated toward an OTT application. As was mentioned before each application has a group of 4 attributes that describe the interaction of the user with a specific OTT application (an example for this case would be Netflix.Flows or Facebook.Flows).

• Application-Name.Flow.Duration.Mean: This type of attributes hold the information related to the mean duration (time) of the flows generated by the user towards a specific OTT application, measured in microseconds. Examples of how this attributes are stored in the dataset are: Amazon.Flow.Duration.Mean or Instagram.Flow.Duration.Mean.

• Application-Name.AVG.Packet.Size: This type of attributes hold the average size of the IP packets that were exchanged in all the flows generated by the user towards a specific OTT application, measured in bytes. It is important to notice that this size is focused on the packet’s header only. Examples of how this attribute are presented on the dataset are: Google_Maps.AVG.Packet.Size or Spotify.AVG.Packet.Size.

• Application-Name.Flow.Bytes.Per.Sec: This type of attributes hold the mean number of bytes per second that were exchanged in the flows generated by the user towards a specific OTT application. Examples of this kind of attributes in the dataset are: Deezer.Flow.Bytes.Per.Sec or Skype.Flow.Bytes.Per.Sec.

• User.Group: This type of attribute represents the objective class of the dataset i.e., the different groups that the users are classified in according to their OTT consumption behavior...