Different types of data analytics

You can divide data analytics in four categories. These four categories are Descriptive, Diagnostic, Predictive and last prescriptive. In this order they are not only increasing in value but also increasing in complexity. Each category not only answers more complex questions but also takes increasingly more knowledge, skills and time to develop the right analysis.

‍

Descriptive

Descriptive analytics is the first level of analytics that only can apply, It very often is also the first step that is done even to just understand what the data is like. With this kind of analysis you get visible how the different values are distributed. The maximums minimums and the distribution over different groups. An example could be to understand the different variables that are their. You could try to bring up the different averages in sales over demographics. Or one could try to understand the distribution in the demographic itself.

‍

Diagnostic

The natural next step after this would be not just to know what the different variables are like but also to go deeper and figure out what they are like. When we do a diagnostic analysis we try often to use more granular drilldowns and timeseries to see why a certain event happens. Among the techniques that can be used to describe what is going on are decision trees. For time series/event logs we could apply process mining. These techniques all give insight into why certain decisions happened. Visualizations can also be a good aid in giving such insight.

‍

Predictive 

Once we figured out why and how certain decisions are made. We can use this not just to understand why things were done and improve on this but we can also predict what will happen in the current setup that there is. We often use the variability in the data to train predictive models and bring forward the relationship between the dependent and responsive variables.

In a world where many things are uncertain or unknown, even a little guidance on what to expect is welcomed. Hence predictive models even when not being able to always perfectly tell you what happens can be a great leverage to gain advantage over your competition or bring that extra insight to your customers needs.

‍

Prescriptive

Now imagine you didn’t just know what will happen in the future but you also know how to change your current behavior such that the future will change for your benefit. A well known example of prescriptive analytics is predicting which machines to service such that you will have the least amount of down time. This means not just predicting if and when a machines will break, but this also means that you have to predict the right action such that you don’t experience any downtime due to broken machinery.

‍

Need help?

Do you need help in determining how far you should push your analytical boundary and what type of analytics you need in order to answer your questions and solve your problem. Then feel free to contact us. We are happy to help you determine the extent of your needs and get you to the forefront of what is possible with analytics.

‍