In keeping with getting to know more about my surroundings, I decided to investigate any significance in the population of New York City. NYC OpenData is a favorable resource when it comes to information about NYC. The search uncovered a summary of the projected population for the years over 2010–2040 by the Department of City Planning in NYC. Besides including totals, the summary sorts the population information into categories of age groups (18 age groups between the ages of 0 and 85+) and by boroughs (Brooklyn, Queens, Manhattan, Bronx, and Staten Island). Time is allotted from the year 2010 through 2040. After gathering examples of types of graphs, refining results, and analyzing the data by way of software, resulted in surprising discoveries. Plotting graphs of the population over time became a way to observe and further analyze patterns in the population summary of NYC in the near future.
Noting how something changes over time is an example of what can be referred to as time-series analysis. Few suggests that “most time-series analysis can and should be accomplished using line graphs” (2009).
When quantitative values are expressed as a series of measures taken across equal intervals of time, this relationship is called a time series. — Stephen Few (2017)
The following are examples of the types of line-graphs that I referenced to help and lead me in plotting the population summary.
Accordingly, heat maps also help to identify particularities in further inspecting the qualitative data (Few, 2009).
Materials and Procedures
The population summary from NYC OpenData came in the form of a spreadsheet that details the per-borough distributions, a further breakdown by age, and the population of each age group as they progress through time. This sheet was further analyzed in OpenRefine, software that allows users to re-categorize the information from the original dataset. It was important to filter the data, to remove things like totals and re-distribute the columns with years into rows under one column labeled ‘Year’.
After parsing the data in OpenRefine, the CSV file was opened in Tableau, where I was able to set the date ranges as time on one axis, and either the age groups or population on the other axis. Tableau also allows the information to be broken down by borough to see distributions based on percentages. By directing the comparison, I led Tableau to produce a heat map and variations of line graphs. The map and graph are presented in a formal dashboard. I was able to produce two sizes, one more suited for viewing on desktop monitors, and the other one especially for mobile and tablet sizes.
This is the desktop version of the dashboard:
To discuss the results, I will complement my discoveries with some of the six basic patterns that are most helpful when analyzing change through time: Trend, Variability, Rate of Change, Co-variation, Cycles, and Exceptions (Few, 2009).
If you study the heat map, it can be noted that the larger population in Brooklyn is mostly due to the six age groups that are below 30 years old when compared to the other boroughs. Queens is very similar. Another important observation through comparison of the overall totals with the by- or per- borough distribution totals, is that you can infer that as Brooklyn, Queens, Manhattan, and Bronx residents age, they tend to leave those boroughs. Though, this data does not tell us where they go exactly, we can note that the 80+ years-old population tends to settle and equalize throughout the boroughs.
When we look at the graphs with the overall totals, whether NYC total or per Borough total distribution, one can observe that the trend in increase in population over time is steady. By looking at these graphs, which are essentially line graphs filled in with the area, you can tell that the rate of change isn’t impactful either. The exception is the five age groups between the ages of 20–44; they cause a spike in the population for certain boroughs (like Brooklyn, Queens, and Manhattan).
This is the mobile version:
Tableau is a flexible program that comes with plentiful options to develop time-series graphs and maps, amongst an unlimited amount of visualizations. To lessen the strong learning curve, I recommend Tableau re-asses their menu and navigation links. Some essential functions tend to be a click away, while others result from user-performed, drag-and-drop actions. There is a top-level navigation that is in-pane, as well as the application’s drop-down menus at the top. I suggest labeling the top-level navigation to support the description of the icons and their associated actions. Some menus are hidden by a right-click when you select specific sheets in a dashboard, or elements in a sheet. Those menus can be transposed and dispersed to the in-pane, side and top menus. There’s no need to hide the important stuff. And with some labeling, users can support their tutorials. Other than that, I am excited to learn more about handling mapping options through plug-ins, and in general, to support more exploration of different types of visualizations.
Department of City Planning (2014, April 22). Projected Population 2010–2040 – Summary. Retrieved July 3, 2018, from https://data.cityofnewyork.us/City-Government/Projected-Population-2010-2040-Summary/ph5g-sr3v
Few, S. (2017). Effectively communicating numbers selecting the best means and manner of display. 2005. URL: https://www. perceptualedge. com/articles/Whitepapers/Communicating_Numbers. pdf [accessed 2017-01-20][WebCite Cache ID 6nfQPuMSP].
Few, S. (2009). Now you see it: Simple visualization techniques for quantitate analysis. Oakland, CA: Analytics Press.
Olson, R. (2016, March 04). Revisiting the vaccine visualizations. Retrieved July 5, 2018, from http://www.randalolson.com/2016/03/04/revisiting-the-vaccine-visualizations/
Roser, M., & Ortiz-Ospina, E. (2017, April). World Population Growth. Retrieved July 5, 2018, from https://ourworldindata.org/world-population-growth