Business intelligence (BI), data analytics, predictive/advanced analytics, data mining, data science, statistics and ML. They are all interrelated and overlap greatly, so it can be difficult to demarcate them. Therefore, different people define them in different ways.
Data analytics and Predictive/advanced analytics
Data analytics refers to the process of inspecting data to extract useful information. There are different types of data analytics and they differ in their goal.
Descriptive analytics tries to describe what has happened. For example, we would use descriptive analytics to answer the question:
“How many parcels did Your Model Car ship last month?”
Predictive analytics tries to predict what is going to happen in the future, e.g. when a customer receives a parcel. It is therefore another type of data analytics, which usually involves the use of ML methods to create models capable of making predictions.
The types of data analytics typically also differ in their degree of complexity. Those analytics cases that use more involved methods are referred to as advanced analytics.
Business intelligence (BI) is the application of data analytics to business information and data. The goal of BI is to provide historical, current and future snapshots of a company and its operations. The methodologies it employs to do that are manifold and include, for example, data visualization, sometimes ML methods, interactive dashboards, and descriptive analytics in the form of reports. For example, a standard BI measure would be to create weekly reports of how many parcels Your Model Car has shipped and how many were returned.
Business intelligence has been practised by many companies for decades already. It became really popular during the 1980s and 1990s. Back then most of the data in companies was stored in relational databases and data warehouses, which store structured data ( 21). That is why BI is usually associated with the analysis of
Today, BI is still popular. However, the term (and discipline) is starting to become somewhat out of vogue, and is being superseded by other buzzwords, notably data science. In fact, data science is often viewed as having evolved from BI because it has the same goal of creating value from company data, but it uses a wider range of
methodologies, including more sophisticated ones that also take on unstructured data ( 21).
Data mining is somewhat of a misnomer. The goal of data mining is to mine for patterns within data, rather than mine for data itself. This is done with methods from statistics, computer science and ML. Like BI, data mining employs descriptive and diagnostic analytics, but it also draws on predictive analytic methods.
The concept of data mining appeared sometime around the 1990s and stemmed from the idea of mining for useful information in companies’ (relational) databases. Today, the term is used less and has been superseded by the term “data science”. One could argue that there is a difference between the two, but by and large they are
Data science is arguably the blurriest concept of them all and it has developed into a kind of umbrella concept that subsumes all others. This shouldn’t come as a surprise, since it refers to the use of scientific methods to extract knowledge and create added value from data. Typically, data science methods are applied in companies.
Data science draws on a number of disciplines, notably statistics, mathematics, business analytics, computer science and ML. Given the broad definition, the methodologies employed in data science span a wide range. These include all types of analytics methods and ML, but also extend to data visualization of both structured and unstructured data. Within companies, the application of data science methods also encompasses the analyses of business processes and data preparation.
Data science really took off in 2012, when Harvard Business Review magazine named data scientist “the sexiest job of the 21st century”.xxiii
Some regard data science as having evolved from BI or data mining, since it is based on a similar idea, but it employs a wider and more progressive range of methods, such as ML. Others with a more critical stance claim that data science is just a rebranding of statistics, BI or data mining.
Statistics is like the mother of all the previously mentioned concepts and disciplines. While all of them are phenomena of the 20th and 21st centuries and are associated with data in digital form, statistics has been around for centuries already. Generally put, statistics is the discipline that deals with the collection, analysis, interpretation and
presentation of numerical data.
Statistics itself draws heavily on mathematical methods such as calculus. Statistical methods are central to all of the a forementioned concepts and disciplines. In fact, ML draws so heavily on statistical methods that a popular saying goes: “Machine learning is essentially a form of applied statistics.”