Diagnostic analytics

1. Diagnostic analytics

In this video, we move on to the second type of analytics: diagnostic analytics.

2. Analytics overview

Just like a doctor diagnosing a patient to find the underlying disease, diagnostic analytics aims to find the root causes of events. It is similar to descriptive analytics but focuses more specifically on the 'why' question.

3. Why use diagnostic analytics?

Diagnostic analytics can be used to find potential causes of events or reasons for behaviors, investigate causal relationships, and, most importantly, the results of a diagnostic analysis can be used to suggest solutions based on the identified causes. While diagnostic analytics can provide evidence for possible causes, it can be tough to prove that there is an actual causal relationship. This is because a statistical relationship between variables can be caused by a third factor we don't know yet or even by coincidence. This is why domain knowledge is a crucial asset in diagnostic analytics, to help assess whether the results are plausible. Additional analysis and setting up experiments can also help strengthen the case for causality.

4. Common techniques

A common technique for diagnostic analytics is drill-down analysis, which consists of moving from general summary data to examining underlying data, for instance, from a sales report showing rapid revenue growth and digging deeper to identify the stores and products that contributed most to the increase. Other common techniques include correlation and regression analysis, which can tell us if variables are closely related; hypothesis testing, when the probable causes are theorized beforehand; and root cause analysis, which we'll discuss further in the next slide. Some of these techniques use similar methods to descriptive analytics, and you will often see similar descriptive statistics and visualizations. The difference is that diagnostic analysis is more directed at underlying causes, while descriptive analytics offers a broader view of an event.

5. Root cause analysis (RCA)

Root Cause Analysis or RCA is a formal set of steps used to look beyond superficial causes that have a direct effect, named 'contributing factors,' by investigating these factors further to identify their causes. This drill-down process continues until the 'root causes' are identified. How far this process goes is typically determined by domain knowledge. An RCA process consists of five steps: first, define the specific event under consideration. Second, collect all relevant data related to this event. Third, determine contributing factors based on an analysis of the data. Fourth, drill down to root causes. Finally, recommend solutions based on the identified root causes.

6. Case study: airline customer satisfaction

Suppose you are working for an airline and are tasked with determining why the customer satisfaction rate is declining. Using diagnostic analytics, you could identify some potential causes by finding and grouping keywords in complaints and ranking the most common categories. You could also investigate connections between the complaint categories and other data you might have like waiting times or delays. This is typically based on domain knowledge. The results of this analysis can then be used to suggest potential solutions for the most common categories of complaints.

7. Let's practice!

Now that you know all about diagnostic analytics, let's practice.

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.