The Four Types of Data Analytics Explained With Examples
Many businesses believe great analytics starts with great data. Sure, data is important, but the best analytics actually start with asking the right questions. This is because each type of data analytics answers a specific question, giving you unique insights into your data.
There are four types of data analytics. Descriptive analytics looks at past data and tells you what happened. Diagnostic analytics goes a step further to explain why it happened. Predictive analytics uses patterns in the data to show what is likely to happen next. Prescriptive analytics suggests what actions you should take to get the best outcome.
Despite serving different purposes, each type has one thing in common. They help you make informed decisions based on what your data says. In the coming sections, we’ll look at each of these analytics types in more detail, along with examples.
Descriptive Analytics
Descriptive analytics is the simplest among the various types. As the name suggests, it simply describes what is happening now or what has happened in the past. It gives you a clear picture of the current state of things: a summarized view of your data through charts, dashboards, or business intelligence (BI) tools.
Descriptive analytics may be enough if all you want to do is monitor key performance indicators (KPIs) and targets like the number of customers, the amount of revenue generated, website visits, or even machine logs in more technical cases. It provides you with just enough insight to observe trends, catch anomalies, and track the metrics needed to make informed day-to-day decisions.
Here’s a real-world example:
Figure 1. Descriptive Analytics Dashboard | Source: HR Power BI Dashboards
This dashboard summarizes hiring metrics, such as the total number of applicants, cost of hiring, revenue generated, and return on investment (ROI) for a recruitment company. It also shows the original source of each applicant (i.e., LinkedIn, referrals, or job boards). This lets the company easily see which sources bring in the most applicants and which ones actually lead to placements and returns.
If the company wants to determine which recruitment channels to keep or stop using, this dashboard gives them just enough insight to make the right decision.
Diagnostic Analytics
Whereas descriptive analytics summarizes events that happened, diagnostic analytics builds on this foundation by revealing the motivations behind those results. Diagnostic analytics explains all the trends, anomalies, and metrics that descriptive analytics brings to light. So, instead of just showing you what happened (e.g., telling you more people than usual visited your website this week), diagnostic analytics digs deeper to show you why it happened.
Data analytics reports often combine both descriptive and diagnostic analytics. Descriptive analytics is used to identify that a problem exists, and diagnostic analytics is used to evaluate the root cause of it.
Diagnostic analytics also involves a lot more one-off analysis because it often answers the question “what changed?”. As a result, analysts have to come up with hypotheses for why the change happened. They do this by collecting the data before and after the change happened, comparing the data, and drawing conclusions.
To get the depth of insight that diagnostic analysis can provide, analysts must use more advanced techniques like drill-downs, correlations, and root-cause analysis (RCA). Businesses use diagnostic analysis to find the source of a problem in order to fix it or to understand what is driving performance so they can double down on what works for their business.
Here’s an example:
Figure 2. Sample Real Estate Dashboard | Source: Real Estate Data Analytics
In the example, a real estate company used descriptive analytics to identify that its revenue and profit decreased in February. Their next step was to use diagnostic analytics to uncover what caused this decrease.
After digging in, they found that the following properties caused a decrease in collected rent because some tenants moved out.
Figure 3. Rental Income by Property Address | Source: Power BI Real Estate Dashboard
Similarly, if you saw that expenses increased, the diagnostic analytics would involve looking at individual expense groups and transactions. Management could then either deem those expenses necessary or look for ways to avoid them in the future.
Predictive Analytics
Once underlying drivers are understood, it’s common for organizations to look ahead. Predictive analytics responds to this need by utilizing patterns to estimate events likely to happen next. Predictive analytics uses patterns in your data to forecast future events. For example, if a store manager wants to know how many sales to expect within a particular month, predictive analytics can give a good guess based on historical data. Predictive analytics involves more advanced techniques like statistical modeling, machine learning, and historical trend analysis.
Businesses use predictive analytics to prepare for the future, mitigate risks, and capitalize on opportunities before their competitors.
Figure 4. Forecasting Sales Trends Using Predictive Analytics | Source: Created by the Author
Apart from forecasting future trends, predictive analytics can be used to predict the likelihood of an event (e.g., customer churn).
The most common predictive analytics techniques are:
- Linear Regression: This assumes a linear relationship between two or more variables and is often used for time-series forecasting.
- Logistic Regression: This returns a value between 0 and 1, which is perfect for predicting the likelihood of events.
- Decision Trees: Used mainly for classification problems, these often consist of a list of yes/no branches that lead to certain outcomes. They are typically used to predict the outcome of a scenario based on multiple variables.
- Neural Networks: These are computing systems inspired by the human brain, which consists of interconnected nodes (neurons) that try to identify patterns and derive a relationship from raw data. Often used when there is no linear relationship between the variables or to check the performance of other predictive models.
Prescriptive Analytics
Prescriptive analytics leverages all the insights from the other types of analytics to recommend the best course of action. It answers the question “Based on everything the data has revealed, what specific steps should we take to achieve the best outcomes?” Prescriptive analytics guides you through the use of optimization models, simulations, and AI-driven recommendations.
Here’s an example of how this type of analytics is used in the real world. A retail company used predictive analytics to find out that demand is going to go up next month. That’s useful, but what do they do about it? That’s where prescriptive analytics comes in. Instead of leaving the manager to guess, prescriptive analytics can tell you:
- How many staff to schedule
- How much stock to order and when
- How much of the budget should be allocated to marketing
In summation, prescriptive analytics gives you clear, calculated, and optimized actions to take so that your data analytics don’t just end at gaining insight but instead lead to tangible business outcomes.
Figure 5. Overview of the Four Types of Data Analytics | Source: Adapted From the Author
Conclusion
These four types of data analytics are not mutually exclusive. Rather, they complement each other. For example, you can use descriptive analytics for explorative data analysis to form your hypothesis. You can then use predictive analytics to test your hypothesis and confirm correlation.
The most effective data analytics teams combine all four types, but this requires a special mix of skills and technology.
Companies that want to invest in descriptive or diagnostic analytics should focus on nurturing spreadsheet data visualizations and BI skills. Those who want to invest in predictive or prescriptive analytics require knowledge of statistics and statistical modelling tools such as R, SPSS, and Stata.
For anyone interested in a more holistic approach, knowledge of data governance and business intelligence infrastructure would make investing in all types of data analytics more effective.
**********
About the Author:
This article was written by Eugene Lebedev, managing director of Vidi Corp, a UK-based data analytics consultancy. He has delivered over 1000 data analytics solutions for more than 600 international clients, including Google, Teleperformance, and Delta Air Lines.
