Ever since the internet hit the mainstream, businesses have been collecting and storing gargantuan volumes of data. Every button clicked, every name entered, and every purchase made are all data – waiting to be exploited.
Google gets at least 3.5 billion searches every day;
Whatsapp users exchange more than 65 billion messages per day;
And, Facebook users collectively share more than 100 terabytes of data per day.
But collecting and storing data alone doesn’t do any good, or let alone makes any sense. Sticking to just collection and storage is similar to having unlimited superpowers and choosing to sleep all day, every day.
Hence enters the four types of data analytics – the four musketeers that can loot information from data sets and deliver actionable insights – assisting companies in making nothing less than data-driven decisions.
If you’re thinking, what are the different types of data analytics and how do they help businesses, then this article is for you. Let’s find out how each type utilises data sets and how their outcomes can help companies to prosper.
The 4 Types of Data Analytics
In essence, the four different types of big data analytics can be summarised as,
- Descriptive analytics – What has happened?
- Diagnostic analytics – Why did it happen?
- Predictive analytics – What is likely to happen?
- Prescriptive analytics – How to make it happen?
To make the comprehension of different types of data analytics easier and memorable for you, let’s consider the analogy of Sherlock Holmes – the world-famous fictional private detective.
As a matter of fact, data analytics can be considered as a case of investigation – the only difference is we’ll have a pool of data to evaluate and not pieces of evidence as in a crime scene.
Let’s take a closer look at the 4 different types of data analytics and how relevant they are for real-world applications.
1. Descriptive Analytics
Descriptive analytics describes what has already happened. It is the most elementary form of big data analytics and can be considered as the starting point of your analytics strategy.
For Sherlock Holmes, descriptive analytics would be similar to looking at the shreds of evidence at a crime scene and explaining to Dr Watson what has happened and just that. Holmes hasn’t started his investigation, and descriptive can be considered as him collecting and presenting the evidence.
Out of the four types of data analytics, descriptive analytics tries to deliver a summary view of what has happened by exploiting historical data (data collected about events that took place in the past). As big data involves multiple data types collected from numerous sources, comprehending such volumes will be tedious or let alone impossible for human brains alone.
This process allows you to describe and present data in a format that doesn’t baffle the non-technical audience. Descriptive analytics starts with the question “what has happened?” and ends there. It is rather fitting to note that the outputs of descriptive analytics are used as inputs for advanced types of data analytics such as diagnostic, predictive, or prescriptive.
With this type of data analytics, companies can determine which among their strategies helped in increasing the sales and revenue and which didn’t – without the answer as to why it did or didn’t.
A simple example of this is A/B testing a paid ads campaign and determining which iteration delivered the best result. Methods such as data aggregation and data mining are used in descriptive analytics to yield useful insights from historical data.
Here are some applications of descriptive analytics.
- Calculating the average duration taken by learners to complete a course.
- Preparing monthly profit and loss statements.
- Measuring the ROI (return on investment) of a social media campaign.
- Inventory tracking at a warehouse.
- Benchmarking yearly sales and revenue of a company.
2. Diagnostic Analytics
Diagnostic analytics answers why did it happen?. It is also called as root cause analysis and can be considered as the successor of descriptive or the next advanced stage of analytics.
Here’s a video explaining diagnostic analytics in simpler terms with an easy to follow example.
With Sherlock Holmes widely being addressed as “the master of deduction”, diagnostic analytics is similar to his conclusion of why a crime was committed by looking at the pieces of evidence and interrogating the witnesses. This will help him uncover the dirty motives of the criminal and determine whether their existence can put anyone in danger.
In this type of data analytics, techniques such as data discovery, data mining, drill-down and correlations are utilised to determine the factors that led to a particular outcome.
As previously mentioned, the results of descriptive are used as inputs for diagnostic. By merely looking at data, analysts wouldn’t be able to answer the questions such as why did a particular store witness increase in foot traffic. With diagnostic analytics, analysts can uncover several hidden stories in data.
The majority of diagnostic analytics solutions use machine learning (ML) and artificial intelligence (AI) to recognise patterns, detect unusual occurrences, identify anomalies, and the factors that contribute to the key performance indicators (KPIs).
In some instances, experts may confine the scope of diagnostic within descriptive analytics itself and thereby discuss 3 types of data analytics.
Here are some applications of diagnostic analytics.
- HR departments can determine how absenteeism or overtime affects the performance of an employee.
- Health organisations can discover the reason why there is a sudden increase in patient inflows.
- Marketers can comprehend why a particular campaign outshined the rest.
Once you have the answers to what has happened and why did it happen, you can move forward with your efforts to answer what is likely to happen.
3. Predictive Analytics
Predictive analytics answers what is likely to happen. Of course, it doesn’t predict any future events per se but helps with predicting the probabilities of an event to take place. In essence, predictive analytics can be compared to fortune-telling minus speculations.
For Holmes, predictive analytics will be similar to predicting when and where to catch Professor Moriarty (the infamous antagonist) by looking at the nature of his past appearances.
The key motive behind predictive analytics is to make data-driven decisions – thereby reducing the risks associated with the “what if we are wrong?” scenario. It gives businesses actionable insights to optimize their marketing efforts, to create products that customers will like using – along with many others.
To make predictive analytics possible, techniques such as data mining, deep learning, statistical analysis, and regression techniques (along with many others) are utilised. This allows companies to predict the probability of an event to occur in the future with the help of historical, transactional and real-time data sets.
The more voluminous and diverse the data sets are, the better. Thus with predictive analytics, the more knowledge you have about the past, the more prepared you can be for the future.
The information gathered from predictive analytics can help companies formulate a set of actions that can positively impact operational effectiveness, sales and profitability, customer satisfaction, branding and many more.
Businesses can identify the customers who are at the verge of churning and offer specialised discounts or incentives to retain them. Companies can also predict customer lifetime value (CLV) and channel their resources to retain customers with the highest ROI.
Here are some applications of predictive analytics.
- Weather forecasting that will help in identifying natural disasters.
- Fraud detection by identifying customer activity out of the ordinary.
- Price optimization to offer competent prices.
- Banks can predict the likeliness of an individual to default on loan repayment.
- Companies can forecast changing trends and tweak their supply chain process.
Once you have the answers to what happened, why did it happen, and what will happen, you can level up and find the set of actions required to answer how to make it happen.
4. Prescriptive Analytics
While other types of data analytics “describe” what, why and how things will happen, prescriptive analytics “prescribes” the actions to take to make something happen.
In the case of Sherlock Holmes, prescriptive analytics can be compared to the actions he suggests to Inspector Lestrade or the Scotland Yard to catch a criminal or to prevent murders. For example, prescriptive analytics in action for Holmes would be suggesting the number of police officers required to catch Moriarty at a particular venue and time.
From a business point of view, prescriptive analytics allows analysts to suggest a course of action that would enable businesses to fully seize an opportunity or eliminate chances of a financial crunch (or even worse, bankruptcy).
Prescriptive analytics heavily relies on artificial intelligence, machine learning, computational modelling, optimization techniques (such as linear, integer and non-linear programming), and rules-based techniques (such as decision trees, inference engines, and scorecards) among many others.
Prescriptive analytics allows businesses to estimate the influence of a particular factor in a potential future.
Google’s self-driving car is an excellent example of leveraging prescriptive analytics. By considering numerous factors such as real-time traffic conditions, upcoming obstacles or potential threats, and route map, the vehicle decides when to slow down or speed up and when and where to make a turn.
Compared to other types of data analytics, prescriptive are least utilised by organisations, primarily because it necessitates advanced technologies and tools to fully embrace – hence expensive and complicated.
However, if you succeed in embracing the potential of prescriptive analytics, it will save you more than you spend on it. If predictive analytics helps a business to determine the likeliness of a customer to churn, prescriptive analytics can deliver a set of actionable insights to retain them.
More precisely, prescriptive analytics looks into what has happened, why it has happened and a handful of what might happen scenarios to suggest the best actions to take.
Here are some applications of prescriptive analytics.
- Multiple route suggestions made by Google Maps considering real-time traffic conditions, distance, and mode of transportation.
- In learning management systems, new courses are suggested that best suit the learner, considering previously interacted content, interests, the pace of learning, and more.
- Helps companies to enhance customer experience and satisfaction by delivering the right products at the right time.
- The healthcare industry can rely on data-driven decisions and recommendations rather than intuition or past experiences.
A data analyst must be keen to observe data without any bias or prejudice – just like Sherlock Holmes. When diving into the pool of data, the analyst must have only a pair of goggles (analytics tools) to see what’s around them and nothing else.
Although many may disagree, data is the new oil and is still in abundance. Businesses that fail to identify the potential of data (more precisely, big data) are more than likely to sink in this data-driven world.
Out of the four types of data analytics discussed, companies may directly jump to predictive or prescriptive analytics (without implementing the four, sequentially) as they may have already established a traditional method similar to descriptive. If you’re planning to become a data analyst, you’ll have to employ one or all of these types.