DataViz Makeover 1 - Singapore Labour Force

A data visualization makeover and a Tableau tutorial for beginners.

Gabriella Pauline Djojosaputro https://www.linkedin.com/in/gabriellapauline/ (School of Information Systems, Singapore Management University)https://smu.edu.sg
01-30-2021

This will be quite a long post, so feel free to jump to the section that you are interested in.

1. Introduction

We are living in a world that is constantly producing a vast amount of data, and data visualisation helps us to understand the data better. However, a poor visualisation can be of little value, or even misleading. Visual analytics is not just about creating a pretty chart, but it is the marriage of data analytics and visualizations that aims to detect the expected and discover the unexpected.

This post will makeover Chart 6: Resident Labour Force by age group from page 22 of Labour Force in Singapore 2019 Report.

2. Original Visualisation Evaluation

To start, we examine the original visualisation from the data (Figure 1) to learn what we can improve. The critiques are given in terms of clarity and aesthetics.

Figure 1. Original chart from Labour Force in Singapore 2019 Report

2.1 Clarity

To keep:

  1. Usage of colours are consistent. All 2009 data uses grey colour, while 2019 data uses blue.
  2. Useful annotations such as reference lines and labels for the median are added.
  3. Source of the data is listed.

To be improved:

  1. Poor choice of chart gives the impression that they are comparing the distribution of age in two categories, for example the age distribution in male and female residents. To represent how values change over time, line chart is a better choice than histogram.
  2. It is difficult to see the values of the data points in the chart. The graph does not have axis and thus we need to depend on the labels to see the actual value of each data point. However, these labels are put below the chart as a table, and not on the chart itself.
  3. The only place where it is mentioned that the values are in percentage is in the subtitle, which is also in an awkward place. It might be better to use the ‘%’ suffix in the value labels.
  4. There is an inconsistency between the statement and the chart. While the statement mentions a decline in 25 - 54 age group as well as an increase in 55 & over age group, the readers are not able to find the values on the chart itself. These are a summation of multiple categories from the chart and the readers must manually sum the numbers together to get the the values mentioned in the statement.
  5. The statement mentioned rise of Labour Force Participation Rate (LFPR) and population size per age group, but the chart also does not have any reference to LFPR or population size per age group.

2.2 Aesthetics

To keep:

  1. The choice of colours are easy on the eye and looks elegant.
  2. The chart has a nice font selection that is easy to read and not unnecessarily embellished.
  3. There is no distracting grid lines.

To be improved:

  1. The chart is not exported in good quality. It is quite pixelated, even in the original report.
  2. Using a very short subtitle ‘Per Cent’ creates an awkward empty space above the chart. It would be better to add it to the title, or omit the subtitle altogether.
  3. Although the report uses a white background, the chart has a grey background colour. Background colour is an unnecessary non data-ink which does not add value in helping to understand the data better.

3. Alternative Graphical Representation

Figure 2 shows the alternative graphical representation proposed for the makeover.

Figure 2. Alternative Graphical Representation

The issues that the alternative design tries to overcome are:

  1. To clearly show the difference of workforce distribution over time by the use of line chart instead of histogram.
  2. To further categorise the age groups together to the same level as the point of interest in the statement, so that the readers can see the value directly without having to calculate the total from different subgroups.
  3. To show the LFPR and population size per age group. This alternative representation will directly show the difference between 2019 and 2009 values instead of showing another line charts because they are just additional information.
  4. To add % suffix in the value labels.
  5. To remove unnecessary non-data-ink by using a white background (Note: The colour on the proposed design is from the paper, which is not going to be used in the final chart).

Additionally, the visualisation will include title and lead-in that highlight the important observations, as well as maintaining the good points from the original charts, such as the choice of fonts and colours.

Update:

Figure 3. Alternative Graphical Representation Update

As the visualisation is being created, there are some slight deviations from the original plan. The changes are as the following:

  1. I could not find a way to remove the gridlines and rotate the y-axis labels if I create a trellis chart. Therefore, I decided to create 2 separate bar charts instead, but I will maintain the order of the age groups.
  2. Annotations added to the line chart to reduce awkward empty spaces.
  3. The two supporting charts are placed directly next to the line chart so it will look more coherent.

The final look of the data visualisation makeover is shown in Figure 4. It is also available in Tableau Public.

Figure 4. Data Visualisation Makeover 1 Final Look

4. Step-by-Step Description

In this section, we are going through the steps to recreate the Data Visualisation Makeover shown in Figure 4 using Tableau. Tableau Desktop has a 14-day trial that can be downloaded here. The steps might be too detailed, but they are designed so that beginners with no Tableau experience can follow the tutorial easily.

4.1 Data Source

There is no visualisation and analytics if there is no data. The data that will be used are:

No Data Source
1. Resident Labour Force by Age Labour Force 2019 Report, Table 7
2. Labour Force Participation Rate by Age Labour Force 2019 Report, Table 5
3. Population Size by Age Singapore Department of Statistics

4.2 Resident Labour Force Line Chart

This chart will show how percentage of each age groups of interest in the Resident Labour Force changes through the course of 2009 to 2019. Steps to create the visualisation is shown below.

Data Preparation

Figure 5. Connect to Data
Figure 6. Add Table
Figure 7. Hide Column

After the cleaning, we still have a column that we are not interested in. There are some values in the column, but it is just an annotation.

The first row in the table has null in the Age column. This value is the total from all Age groups, which we can obtain with the SUM function later. We can filter out this row.

Figure 8. Add Filters
Figure 9. Select Age

We can either select None and manually check the data items we want one-by-one, or check all boxes first and unselect the one we do not want. Since we only want to remove one age group, we will do the latter.

The table is currently in a wide format, but we want it to be long and thin so we can assign the years as a Date variable. We can use the pivot function from Tableau.

Figure 10. Pivot
Figure 11. Rename Columns

The data is now transformed to a thin table with three columns. The original column headers are stored in Pivot Field Names, while the values are stored in Pivot Field Values. Rename the columns to make it easier to understand.

The original and new column names are shown below.

Original Column Name New Column Name
Pivot Field Names Year
Pivot Field Values Resident Labour Force
Age (Years) 2 Age
Figure 12. Change to Date

Notice that there is an ‘Abc’ symbol above the column Age and Year. This means the two variables are considered String Variables. However, we want the Year to have a Date data type. * Click the ‘Abc’ symbol in Year column header > Date.

Figure 13. Create Group

Next, one of the improvement we want to make on the original visualisation is to group the age variable so it will be the same as the age groups that are mentioned in the statement. Therefore, we need to group into 25 - 54 and 55 & over. Age group 15 - 24 will be created even though it is not mentioned in the statement to show that the numbers sum to 100%.

Figure 14. Age Groups
Figure 15. Create Calculated Field

The current value that we have for Resident Labour Force is absolute value, but we are interested in the distribution of age in the workforce. We can get this value by doing a quick table calculation or creating calculated field. This post will use the latter method.We need to create two calculated fields: Total Resident Labour Force and Percentage of Total Resident Labour Force.

Figure 16. Total Resident Labour Force Formula

The formula shown above calculates the total number of people in the Resident Labour Force per year.

[Resident Labour Force]/[Total Resident Labour Force]

Figure 17. Go to Sheet

We are done! We can go to the Sheet1 Tab in the bottom left-hand corner.

Creating Visualisation

Sheet is where we can build visualisation. The general layout in Tableau interface is shown in Figure 16.

Figure 18. Tableau Layout

We can build visualisation by dragging the fields from the data pane directly to the canvast or to the shelves. Depending on where we drop the fields, they will take on different roles in the visualisation. To create the line chart for the Resident Labour Force by age group, we will drag the following fields to the column.

Figure 19. Rows and Columns for Line Chart

Notice that Tableau automatically recommends a line chart to visualise our data. Don’t be alarmed that it is only displaying a straight line. Right now it computes the total percentage of all group, that is why it is always 100%. Next, we need to separate the age groups.

Figure 20. Add Age Group

Customisation

We can further customise the chart appearance to enhance the usefulness and readability of the chart. There are several customisations on the line chart that we are going to perform.

Label

Figure 21. Add Label
Figure 22. Format Label

Axis / Header

As we have already labeled the data points of interest, the y-axis is no longer necessary. We can remove the axis to increase our data-ink ratio.

Figure 23. Remove Axis
Figure 24. Hide Field Labels for Columns

Line Markers

Figure 25. Add Markers

Title

Our current default title is the sheet name, which is Sheet 1. We can change the sheet name, and further tailor the title to make it more informative.

Figure 26. Rename Sheet Name
Figure 27. Change Sheet Title

‘’ means the value is dynamic. Whenever the sheet name changes, the title will follow accordingly. We can also add other variables by clicking Insert. However, we are not adding other variables now.

Annotations

We can also annotate on the chart itself to highlight interesting findings.

Figure 28. Annotate Mark
Figure 29. Annotation

Just like the sheet title, we can customise the annotation and make use of the variables using the Insert button. And now, we are done for this line chart! We can adjust the width and height of the chart by hovering to the right side or below the chart until our cursor turns into a double-sided arrow and drag accordingly.

Figure 30. Final Look for Singapore Resident Labour Force Line Chart

4.3. Labour Force Participation Rate Bar Chart

Next, the statement mentions that the increase in older residents in Singapore Labour Force is due to the rise in Labour Force Participation Rate for older residents. We are going to visualise this data to support the statement. Some steps are the same as what was explained in the previous section.

Data Preparation

Figure 29. Add New Data Source
Figure 31. Add Filters

If we scroll down and see the bottom of the data table, we can see that they have already grouped the age. Therefore, we can filter out the 5-year age bins on the top.

Figure 32. Live Connection

The spacing seem to be very wide for the age groups. Since we are using a live connection, any update on the data source will be reflected in Tableau. I opened the Excel file to manually fix the spacing.

Figure 33. Labour Force Participation Rate Percentage Change Formula

Although this data is also in a wide format, we only need the difference between 2019 and 2009 data. We do not need to pivot the table, but we need to create a calculated field for the difference.

Figure 34. Add New Worksheet

We are done preparing the data. Now we can create the second chart on a new worksheet.

Create Visualisation

Figure 35. Rows and Columns for Labour Force Participation Rate Bar Chart

For the bar chart, we want to have 3 bars that show the change of Labour Force Participation Rate between 2009 and 2019.

Customisation

First, we are going to add the colour and label to make the bar chart easier to see.

Figure 36. Add Colour and Label

Colour

If we see the color legend on the right, we can see that the range does not start from 0, it only uses the smallest and the biggest value. The purpose of changing the colour to to quickly highlight the positive and negative values as well as showing the significance of the change. Colour is a pre-attentive attribute, so we can quickly notice the difference in colour.

Figure 37. Change Colour

The colour of the bars may not change now because all the values are positive. It is okay.

Label

Even though the Labour Force Participation Rate % Change is a percentage, the number is not between 0 and 1. Therefore, we will use a slightly different method to display the ‘%’ sign.

Figure 38. Customise Label

Now that the bar is labeled, we can take away the unnecessary axis and headers using the same steps as in the previous section. Remember to rename the sheet name and put a proper title.

Figure 39. Final Look for Labour Force Participation Rate Change

4.4. Population Size Change Bar Chart

The steps to create the Population Size Change Bar Chart is exactly the same as the Labour Force, except we are using a different dataset and we are not displaying ‘%’ sign in the suffix.

Figure 40. Final Look for Population Size Change Bar Chart

4.5. Dashboard

Now, it is time to combine the various charts that we have created.

Figure 41. New Dashboard
Figure 42. Dashboard Size
Figure 43. Insert Sheet to Dashboard

By default, Tableau uses a Tiled layout. To better arrange our sheet, we can make use of the objects.

Figure 44. Vertical Object

Vertical object allows us to have a vertical grid that we an fill in with sheets or another objects.

Similarly, we can use blank to add paddings around the objects (such as to align the sheets), add horizontal grid, text, etc.

Figure 45. Remove Legend

Since we have annotated the line chart and the colour in the bar charts are to highlight positive and negative values, the legends are not necessary.

Figure 46. Show Dashboard Title

To create a title and lead-in for the dashboard, we can either use a Text object or use the dashboard title itself. The Show dashboard title checkbox highlighted in Figure 45 will make the dashboard title appear.

Figure 47. Final Dashboard Layout

The final dashboard that I created is shown in Figure 46, marked with colour-coded rectangles to show what object I used to arrange them.

Congratulations, we have completed our dashboard!

5. Major Observations

Figure 48. Singapore Resident Labour Force Alternative Graphical Representation
  1. Due to the rapidly rising population size of older residents, having more older residents in the workforce is natural, or even preferable. As the distribution of population is getting more and more skewed, Singapore’s old-age support ratio will continue to decrease. Old-age support ratio is the number of people who are financially productive compared to the number of older people who may be dependent on others’ support (Singapore Department of Statistics). The increase of Labour Force Participation Rate in older residents helps to reduce the number of older residents that depend on others’ support.
  2. The increase in population size in 55 & Over age group is incomparable to the decrease in population size of 25 - 54 age group. This means that the majority of people in the 55 & Over age group in 2009 are still alive in 2019, which can be attributed to Singapore’s excellent healthcare system. We will continue to see rising number of older residents in the workforce before the ratio will reach an equilibrium as long as the population size of younger residents does not continue to decline.
  3. As the increase in Labour Force Participation Rate is not as high in 15 - 24 and 25 - 54 age groups compared to the 55 & Over age group, another way to sustain the old-age support ratio is to increase the Labour Force Participation Rate in the younger residents. Non-working residents in the younger age groups are most likely students or stay-at-home mother. Finding out the reason they choose not to participate in the labour force is a necessary step if we want to encourage them to enter the workforce.

Citation

For attribution, please cite this work as

Djojosaputro (2021, Jan. 30). Gabriella Pauline: DataViz Makeover 1 - Singapore Labour Force. Retrieved from https://gabriellapauline.netlify.app/posts/2021-01-29-dataviz-makeover-1/

BibTeX citation

@misc{djojosaputro2021dataviz,
  author = {Djojosaputro, Gabriella Pauline},
  title = {Gabriella Pauline: DataViz Makeover 1 - Singapore Labour Force},
  url = {https://gabriellapauline.netlify.app/posts/2021-01-29-dataviz-makeover-1/},
  year = {2021}
}