Application

Facebook uses the recommended system to suggest friends, show products the user may like, and videos, to name a few, using data analytics and big data. According to the Facebook website, they use common friends, locations, Facebook groups, photo tags, and personal contacts were uploaded to identify the suggestions to recommend friends may know (Meta, n.d.).

Current Approach

Facebook is the biggest social media platform, with 2.91 billion active users in the last quarter of 2021 (Statista, 2022). The data that Facebook used is conceded big for many reasons. First, they have huge amounts of data (volume) to store and analyze, more than four petabytes per day (Roy, 2021). In addition, Facebook analyzes everything when users have been active on the website, starting with the pages visited and device information to the mouse motion on the screen (Ganjoo, 2018). In the matter of data velocity, as the users have been active on Facebook, they generate data. According to eMarketer Editors, users spent 34 minutes daily on Facebook on average (2020). This means users spend 16.49 million hours on Facebook per day. Furthermore, around 40 thousand queries are done on Facebook each second (Roy, 2021). Likewise, around 510 thousand posts are published per minute (Heitman, 2021). All these numbers proved a huge data velocity that Facebook needs to process and analyze.

In addition, Facebook supports multiple types (variety) of data to be posted and received data from multiple sources. For example, text, images, videos, and locations are examples of content that users can post. Each kind of these data type has its way to analysis. For example, Facebook has algorithms to analyze images to identify the persons in photos and suggest tagging them.

Medicine and Health Use Case: Forecast Admission Rates

Application

Hospitals use big data to perform productive analyses to forecast admission rates to be prepared and provide efficient patient service.

Current Approach

Using multiple data sources, internal and external data, hospitals apply predictive analysis to estimate the patient’s admission rate (Cprime Studios, 2022). The internal data include current patients’ data in the hospital and historical data. The external data include the weather conditions, natural disasters, and emergencies in the area. All these data are considered big, representing the volume of big data characteristics. In 2021 the patient’s file volume in the United States had been around 2765 GB. According to Suter-Crazzolara, each patient’s file size takes up to 80 megabytes per year for each patient. Also, there were more than 36 million patients in the United States in 2021 (American Hospital Association, n.d.). In addition, medical data is big data because it contains multiple data types. For example, there are doctor notes, lap results, X-Ray images, and sometimes 3D images for each patient. These different kinds of data represent the data variety in the hospital data, making it complex. Also, capturing the external data in real-time frequently reflects the velocity, which is also one of the big data characteristics, requires high network speed and machines to process and analyze those data.

Scientific research Use Case: Space scientific discoveries

Application

Exploring space is one of the most important types of science. The Square Kilometre Array (SKA) project uses big data to discover space efficiently.

Current Approach

Thousands of dishes and up to a million low-frequency antennas will ultimately be used in the SKA, allowing astronomers to watch the sky in unprecedented detail and sweep the whole sky far quicker than any other system now in use (SKA, 2020). SKA project generates five zettabytes (106 petabytes) yearly (Pool, 2020). This high volume of data shows how big data could impact the scientific research field. The data generated from the dishes come from different sources with different formats representing the variety of data. Around 14 countries involved in this project are Australia, Canada, China, France, Germany, India, etc. In addition, these data need a high machine that can process the high data velocity. The database receives real-time data from antennas spread over the world, reflecting the velocity of data received.

Sales/Marketing Use Cases: E-commerce Personalize Page

Application

E-commerce can use the products and users’ data to increase their profits by customizing the page based on user interest. Increasing the profits can be done by analyzing the historical user data and the products.

Current Approach

Each product has many characteristics, such as colours, size, and weight, to name a few. Also, each user has interests that can be known by analyzing their favourite books, movies, and products. E-commerce websites can use this data to show the products that fit the customer’s needs by analyzing the history of purchases and recent products viewed. For example, if a customer likes technology gears with blue colour, the e-commerce can show all technology products that have a blue colour and show the image that shows the favourite colour. To perform personalization, e-commerce page companies need to analyze huge amounts of data and apply complicated algorithms. Amazon, for example, has 300 million active users in 14 countries (Petrov, 2022) with more than 12 million products (Dayton, 2022). All these numbers generate a huge amount of data, which represents how high volume of data. Also, Amazon collects data from different sources in many ways that users can visit Amazon, such as web browser, phone app, Amazon music app, Amazon prime video, and Kindle store. All these sources create, present and process a variety kind of data, such as images, text, videos, and device information, to name a few. Furthermore, Amazon, on average, sells more than 240,000 products per hour, which represents the high velocity that Amazon servers handle.

Politics Use Case: Election Forecast

Application

According to an article posted in http://analyticsinsight.net/, from the early 2000s, big data became an essential role in predicting the election result (2021).

Current Approach

Using big data to forecast election is not easy and require high expertise and data accessibility. Agencies that take responsibility for election campaigns use big data to study the factors that lead to the success of their campaign by analyzing high-volume data from different resources every day to estimate the campaign’s chance of success or failure. Social media is a valuable source of data that can be analyzed and results from each second by analyzing people’s behaviour. In 2016 Trump’s digital media director said, “Facebook and Twitter were the reason we won this thing. Twitter for Mr Trump. And Facebook for fundraising” (Lapowsky, 2016). The number of tweets related to the presidential campaign in the United States of America in 2016 hit more than 1 billion tweets. This is a large amount of data to analyze. Also, to analyze these tweets, developers need to analyze the tweet’s location and date/time and apply NLP to test the tweet. This complex variety of data requires advanced software and hardware to process the data. In addition, the analytical team needs to process the data in real-time, which provides more accurate results, indicating the high velocity they deal with.