Breaking Down the Big Three: Data Science, Big Data and Data Analytics

Data. This four-letter word has unleashed a new industrial revolution and unprecedented changes in our lives, transforming how we do old things and enabling us to do new things that we had never thought possible. While we live in the data age, many people still need help distinguishing the differences between data science, big data and data analytics. Are you currently enrolled in a Master Data Science online course? Or are you just simply data-curious? Keep reading to learn about the major differences between these three key areas, and for you to better understand the technology that influences our everyday lives. 

 

Data Science

 

One should view data science as the building block for big data and data analytics. Data science generally refers to the multidisciplinary academic field that uses scientific methods, processes, systems and algorithms to extract knowledge from various information sources. The field combines statistics, mathematics, programming and problem-solving to obtain, process and eventually allow analysis for insights. 

 

A data scientist, ranked as ‘the sexiest job of the 21st Century’ by Havard Business Review, prepares and creates algorithms and models to collect and organise raw information into a dataset. They are also responsible for other practices such as data cleansing which is the process of identifying and rectifying corrupt or inaccurate data records in a set. 

 

In data science, one usually uses programming languages like Python, R and Julia to create these algorithms. It covers anything related to data and can be summarised into four clear stages:

 

• Data ingestion – the collection of data, which includes ‘big data’. 

• Data storage and processing – processing and managing that data into a broad range of systems and datasets. 

• Data analysis – analysing the data. 

• Communication – publishing this analysis into reports, graphs or other mediums. 

 

Data science has been key for enterprises and institutions to improve outcomes, empower decision making and solve problems more effectively. It can be found anywhere, everywhere across all industries. 

 

Big Data

 

Contrastively, big data refers to giant datasets that cannot be processed under traditional means (e.g. application software). You may have heard it be used loosely across the media, such as during the infamous Cambridge Analytica-Facebook scandal where data was exploited for political motives in the U.S. – however, big data is not a monolith. It is ubiquitous and heterogeneous at the same time. Three forms of data structures constitute what we refer to as ‘big data’: 

 

• Structured Data

 

This refers to any data that is recorded in a fixed and standardised format, is easily accessible and can be processed. From the standpoint of big data, structured data is the best to work with for obvious reasons, but also because they function through predefined formats and have highly coordinated measurements. Structured data includes birthdays, addresses, age, credit card numbers, contact numbers and dates. The sources of structured data can vary significantly. 

 

For users, they include, but are not exclusive to any response made from surveys, click-stream data that records the actions a user makes on their website, move-by-move breakdown of actions taken in a game, and any data inputted by a user onto a ‘form’ (e.g. a hotel reservation). There is also structured data generated from the computer or machine itself, including sensor data, web logs and financial information. 

 

• Unstructured Data

 

As per its name, unstructured data refers to datasets not stored in any predetermined or structured format. They are also called ‘modern data’ given their existence is inextricably linked to the digital age. To get a picture of what this means, unstructured data is stored in a ‘digital wilderness’, and these include Word documents, PDF files, HTML files, images such as JPEG, PNG and GIF, audio files such as MP3, WAV and FLAC, video files such as MP4, AVI and MOV, system log files, and most commonly, social media content such as comments, posts, likes, shares etc. 

 

As unstructured data lacks order and schema and is far larger, they are comparably more challenging to work with than their structured counterpart. Yet today it forms a significant bulk of big data overall, accounting for 80 to 90 per cent of all forms of data and having a growth rate three times faster than structured data. Unlike structured data, unstructured data provides businesses with far more nuanced and granular insights, meaning that enterprises can make more qualitative assessments. 

 

• Semi-Structured Data

 

Datasets that are consistent and have a degree of organisation, but do not fit into a rigid data model, are called semi-structured data. Usually, this can include email messages, webpages, system log files, EDIs (electronic data interchanges which are computer-to-computer transmissions of things like purchase orders, invoices etc.), CSV, XML and JSON (languages used to communicate and transmit data from a web server to a client’s device), as well as log files. 

 

Semistructured datasets cannot be stored in traditional databases but are instead grouped based on metadata and tags. While only representing about 5 to 10 per cent of the information that modern businesses deal with regularly, they have been growing steadily with the rise of artificial intelligence (AI) and machine learning (ML). 

 

Enterprises and institutions use big data to monitor and predict traffic. Big data becomes even more important for AI and ML as it serves as the lifeblood for both technologies and the advancement of their algorithms. Making big data work, however, requires an adequate strategy and architecture. Specialised platforms such as Snowflake, and Databriks, or cloud options such as Azure provide effective solutions for big data management. 

 

Data Analytics

 

Finally, data analytics is essentially placing the previous two into practice. The data analyst is responsible for filtering large swathes of big data and extracting the most relevant information from it to understand trends, patterns or variance. They are like ‘intelligence officers’ since they use this ‘intel’ to develop answers to questions posed by businesses and institutions. There are four main types of data analysis methods that analysts use:

 

• Descriptive Analysis – understand what has happened and is happening in the dataset. 

• Diagnostic Analysis –  understand the causes of that event.

• Predictive Analysis – predict what will happen in the future through the dataset. 

• Prescriptive Analysis – predict how this predicted event will happen or unfold. 

 

An e-commerce company for example would use data analysis to measure a customer’s preferences and monitor buying behaviour. The information can personalise customer experience, make sale targets and opt for more effective marketing strategies. In fields like e-commerce, data analytics becomes particularly important for business success.

 

However, data analytics is not always about the customer, as it can also be used to garner insights into internal operations. Things like risk management today go hand in hand with data analytics, and it can be used as a valuable tool to identify, interpret and analyse internal risk to ensure that enterprises make informed decisions. Without data and timely analysis, one would be unprepared for uncertainty, hindering long-term success. 

 

Data science, big data and data analytics will be integral parts of the future of business and industry. Data penetrates almost every aspect of our lives today, so it is important to equip yourself to understand their differences.