Data analytics is the science of analysing raw data in order to derive useful insights and can be classified into four types – Descriptive, Diagnostic, Predictive and Prescriptive. No one type is better than the other, and businesses are most successful when they use them in tandem.
In this blog we describe the differences between Descriptive, Diagnostic, Predictive and Prescriptive Analytics.
Descriptive analytics is a technique that summarises raw, historic data from multiple sources, in order to provide insights into the past and help answer the question of what has happened. This allows us to learn from past events and outcomes, so we can make more informed decisions for the future. Highly data-driven organisations do not solely rely on Descriptive Analytics and find it more effective combining other types of analytics to explain why certain data trends appear.
Most of us use Descriptive Analytics every day. A marketing department will use Descriptive Analytics (in the form of Google Analytics) to learn about their website visitors, whereas a finance department will use Descriptive Analytics to break down costs per product over a period of time.
Diagnostic Analytics examines historical data from different periods of time, in order to answer the question of why something has happened. This type of data analytics enables us to identify patterns, trends and correlations to help us uncover the root cause behind a problem. This can be done using data mining techniques like Regression analysis, Anomaly detection, Clustering analysis and Detection analysis.
Typical examples include IT departments that use Diagnostic Analytics for troubleshooting, as well as customer service centres that use Diagnostic Analytics to find the root cause behind certain customer behaviours.
Predictive Analytics uses data from Descriptive and Diagnostic Analytics, and applies statistical models to make predictions for future events. This enables organisations to estimate the likelihood of different outcomes, identify risks and opportunities, and automate decision making processes based on historic data trends. Although Predictive Analytics helps us to understand what is likely to happen, this is only an estimation the accuracy of which depends on the quality of the data.
Amazon is one of the most successful examples, using Predictive Analytics to power its recommendation engine, which once you have purchased a product from Amazon’s website, it will recommend other products that customers with similar purchasing patterns have opted for. Another use that we have all benefited from is weather forecasting; thanks to Predictive Analytics, in 2019 we are able to successfully predict the weather 10 days into the future compared to the 1980s when we could only forecast one day in advance.
Prescriptive Analytics helps organisations decide on what action to take to solve a problem based on outcomes from similar events. Prescriptive analytics requires both Descriptive and Predictive Analytics. While Descriptive Analytics aims to provide insights into the past and Predictive Analytics estimates what might happen, Prescriptive Analytics determines the best solution or course of action among various options, given the known parameters. Machine Learning is one of the Artificial Intelligence techniques used within Prescriptive Analytics to continuously process new data and adjust the algorithms and corresponding results automatically.
Prescriptive analytics can help prevent fraud, limit risk, increase efficiency, meet business goals and create more loyal customers. Airlines use Prescriptive Analytics to automate the decision-making process to optimise ticket pricing. Many factors can affect ticket prices, and, to maximise profits, they need to consider customer demand, competition, oil prices, weather and time of year. Prescriptive Analytics will suggest the optimum pricing policy based on the fluctuations of the afore mentioned factors at any given point in time.
Final thoughts
While data analytics can be simple, as data proliferate, its analysis brings computational and data-handling challenges. This is where big data-capable systems and skilled data professionals like data engineers and data scientists can help organisations make the most out of their data.
