Data analytics is the process of using data to answer questions about the information it represents.
Databases don’t just store data; they also make that data accessible for the purposes of creating reports and performing data analytics. By analysing data stored in databases, we can make better decisions about what to do in the future.
Most organisations will use a database to keep track of useful statistics about their performance.
Data analytics can be split into four types:
- Descriptive analytics: reporting on what happened
- Diagnostics analytics: analysing why things happened
- Predictive analytics: predicting what might happen
- Prescriptive analytics: deciding what should be done
Descriptive analytics describes what has happened, and is often the process that’s used for creating and tracking progress against key performance indicators (KPIs), metrics, and reports.
The outputs are typically numerical values that answer a defined question, for example:
- Student attendance percentage
- Monetary value of sales
- Number of appointments made
The analytics produced are often for a particular time period, for example ‘Sales in the last month’ or ‘Average grade this year’.
An example from the computer sales database would be a report that shows that sales of computers are up, but profit is down.
Diagnostic analytics involves looking at the findings from descriptive analytics to attempt to find out why certain things have happened. This may involve asking questions such as, ‘Why have students numbers dropped this year?’ or ‘What caused an increase in website traffic?’
This diagnostic process can be broken down into these three steps:
- Identifying unexpected changes or anomalies in the data
- Collecting additional data about these discrepancies
- Finding trends or links that describe why the anomaly occurred
Often, additional data such as the dates of public holidays, weather reports, or birth rates, is needed to identify the reasons that something happened.
An example would be using data to investigate why profit is down, and finding that the stores’ rental costs have increased.
Predictive analytics looks to answer questions about what might happen in the future.
Typically, historical data is used to find trends or identify patterns. Possible uses for this include:
- Predicting sales
- Planning for data storage growth
This form of analysis is also used to answer ‘what if’ questions such as, ‘How will sales be affected if we have a hot summer?’ or ‘What will the effect be if the cost of raw materials increases?’
For example, in the computer sales database, you could predict that if rental costs remain the same, the company will have to sell 10% more computers.
Prescriptive analytics looks to answer questions about what should be done. The output from this analysis identifies the likelihood of outcomes if particular actions are taken or particular events occur.
The likelihood of different outcomes can be estimated if you look at what happened in the past, for example:
- If the birth rate this year is 5% above the average, there is an 80% chance the school will need to hire another teacher in five years’ time
- If the weather forecast suggests snow for more than five days in a month, there is a 50% chance at least one snowplough will need to be repaired
Prescriptive analytics could help you to construct a scenario such as, ‘If the company moves and reduces rent by 20%, profit will increase by 17%.’
The field of data analytics is a complex one; it has many different terms and descriptions, which often mean different things to different people. The glossary at datascienceglossary.org covers most terms, with links to resources to find more information.
- Is there any way in which data analytics may be useful in your life?
- Do you have any questions about the joins quiz in the previous step?
Share your answers in the comments section.
In the next step, you will learn how to use SQL to perform data analytics.