Make your data insight impactful with the right comparisons

In Data Analytics, especially Data Visualisation, comparisons are crucial to make sense of the data. It would have sounded quite logical like 'duh' but at most times, we underestimate the value of comparisons. Let us consider 2 examples to show this importance:

Example 1

If you go to a foreign country, and wanted to know how much an average person earns in that country, and someone told you $3,500sgd gross salary per month. What does that tell you? Is that a lot or not much? How would you know? You naturally use yourself or the average salary of your country to compare. In the latest statistic in Singapore from MOM in Jan 24 (Summary Table: Income (mom.gov.sg), it says that the median salary for full-time employees including CPF contributions is 5,197sgd. If that is the rough comparison, then you would say that it is less than Singapore. Notice, that is just a rough gauge as comparing median and average (mean) is quite statistically different, which we will cover in another article.

In fact, if you refer to a Numbeo report (Asia: Rankings by Country of Average Monthly Net Salary (After Tax) (Salaries And Financing) (numbeo.com)) comparing the average monthly net salary, you will also see that Singapore's numbers of 6,198sgd is about 1,000 difference from the median salary for Singapore. 

In comparison, it is important to compare apples-to-apples e.g. all in SGD, and the period is the same. In this case, the Numbeo report does not have a date to it, so that needed to be verified as well.

Example 2

Usually in my class, they are given a Singapore Population dataset to which they are to explore and develop findings about it. This data is usually obtained from websites such as Singstats (Singapore Department of Statistics (DOS) | SingStat Website)

Among the questions the participants like to explore, One that would normally surface is to compare town population and the distribution of races in each town. One of the first chart would be to see which town has the highest population. As you can see in the chart, below it makes a lot of sense to compare the population among the towns in Singapore and the findings show that Bedok town has the highest at 41K. Now, you would say that that is indeed correct, and I agree with that finding.

Naturally, we would like to drill further down into Bedok and try to find out the race distribution within Bedok. Here is a chart that could tell us that detail:

From the chart, most participants would say, there you go, Bedok has the most Chinese at 72%. We cannot say they are wrong, right? But think again. For anyone who knows about Singapore, or have lived here would know that Chinese form a majority of the population hovering around 70%. Do we really need this chart to tell us so? Now, how else can we make a comparison to tell us a better answer. Here is my suggestion: run a report of the overall national average distribution of the entire Singapore for Chinese and compare that with Bedok’s distribution.

Here is a report of the national average race distribution and compare it side-by-side with Bedok’s race distribution:

Now, we can derive a richer insight and safely say that even though Bedok has a high percentage of Chinese at 72%, it is still slightly lower than the national average of 74%.

In conclusion, I hope you see the importance of comparisons in order to get a deeper insight into the data to understand what is the situation, and with those insights draw conclusions as well as form decisions based on those insights.