Big data. Artificial intelligence. Machine learning. You hear these terms more and more nowadays, typically in relation to the work that data analysts or data scientists do. And you might be a bit confused by what it all means. But with the advance of data-driven decision making, it’s more critical than ever to understand how data scientists work and how they think.
Artificial Intelligence (AI) gets thrown around a lot and has lots of different meanings to lots of different people. But AI really is an umbrella term for a collection of analytic methods used by data scientists to solve practical decision problems. It doesn’t involve robots (well, not like you see in the movies) or extraterrestrial involvement – it’s simply an analytic method that utilizes data to provide data scientists with a quantitative reason for making a decision.
Using analytic methods based on statistics and data goes back to World War II, where the British had over 1,000 people developing quantitative bases for operational decisions. Since the advent of the computer, this process has obviously been streamlined to an exponential degree.
Today when solving a problem, data scientists – just like reporters, detectives, and doctors – have to understand the question at hand and then explore what information is available to get started. The goal is to consolidate it down to a well-defined business question built around supporting data.
The term Business Intelligence (BI) – just like Artificial Intelligence – gets thrown around a lot and tends to have different definitions depending on who you ask. However, BI simply refers to the algorithm- and model-based tools that are used to help solve the problem or challenge identified in the well-defined business question. It utilizes tools like flowcharts, dashboards, reports, and the more typical charts and graphs. As technology evolves, the tools of BI evolve as well. They help to visualize results to end-users.
Where Business Intelligence is driven by algorithms and models, Machine Learning is purely fueled by data without the involvement of human interaction. It’s ideal for scenarios where a business process is not very well understood as it can analyze data to discover patterns and search for structures. This is where the learning part comes in as Machine Learning involves identifying patterns and structures to train itself to make future predictions and decisions.
This comes in handy in scenarios where you might have large sets of historical data regarding sales, product supply, opportunity creation, prices, and outcomes, but no idea why customers are choosing whether to buy or not. The reason? You can’t formulate it into a mathematical optimization problem. However, Machine Learning can find patterns and make predictions about future opportunities.
Deep Learning picks up where Machine Learning leaves off, teaching machines to perform tasks that normally would require human intelligence.
Deep Learning typically involves a network of technologies that deliver accurate object detection, speech recognition, and language translation. This network is called a “neural network” as it resembles the structure of neurons in the brain, with layers of connections that can learn from data and recognize patterns, classify data, and predict events.
Today’s data scientist has a wide array of technological tools at their fingertips each with a very important role. But there is no “silver bullet” in today’s world of data-driven business. Every method has its benefit and excels in its own way but it takes the mind of the data scientist to take these methods and the information, data, and conclusions they provide to develop business solutions that best fit the need and budget.