Contest: Data Analysis and Visualization

PLEASE DO NOT OPEN UNTIL INSTRUCTED TO DO SO

IMPORTANT:

Any solution received that includes any IDENTIFYING information will not be eligible for
anything beyond “Honorable Mention.”
Examples include any of the following:
• Names of the individuals or school names in the solution
• Names of the individuals or school in the “properties” of the file
• Each program, Adobe, Microsoft, etc. contains metadata about the
document. Remove ALL of these from your document. If you do not know how
please ask someone!
Any solution that does not provide accurate citing of professional resources will be removed
from consideration.

Examples include any of the following:

• Copying and pasting diagrams and images from a website

• Using descriptions and product data verbatim from source
• When in doubt, cite your source
Data Analysis and Visualization Contest
Contest Statement:

Common techniques in data analytics and visualization include data acquisition, data cleaning, handling
missing values, wrangling, data integration, simple data analysis (e.g., outlier detection, identifying
collinearity between variables, dimension reduction, summary statistics, and determining skewed
attributes), statistical model building, finding patterns, finding clusters, plotting geo-temporal points on a
map, data visualization, visual analytics, problem solving, making recommendations, and providing
quantitative analysis to support decision making.

Each team may consist of one or two individuals and must have at least one computer to complete your
project. The contest is written from the perspective of guiding someone who is new to using Tableau.
Appendix A contains a quick-start guide to Tableau’s interface. You are welcome to use other software
and skip the Tableau focused tutorial instructions found in Appendix B – you are only required to turn in
your answers and a visualization (using any software of your choice) for the required tasks.

Contest - Deliverables
Solve the tasks in an individual or two-person team – sharing completed solutions with other teams is not
allowed. You are welcome to utilize existing software to manipulate/analyze data, write your own code,
or follow any other analytical methodology you prefer to arrive at a solution to each task. To complete
this challenge, create and turn in a document that answers as many tasks as you can. Also, turn in any
other supporting files such as computing code (e.g., Python scripts), Tableau workbooks (.twbx file),
intermediate datasets (e.g., CSV files), etc. to showcase your work. Do not upload any part of this contest
or your deliverables to online locations (e.g., GitHub).

Resources - Software recommendations

It is recommended to download Tableau Desktop (free year-long license) or else use online Tableau
Cloud for the contest (see: https://www.tableau.com/academic/students). Alternative analytics or
visualization software that students may find useful include Excel, R, Python, SAS, SPSS, Weka, and/or
Power BI. Students are encouraged to use any software and online resources they wish, including Google,
to learn and complete the challenges.

If you choose to use Python for this contest, SciPy is a set of widely used packages for managing,
analyzing, and visualization large-scale content. It consists of libraries for data structures such as
DataFrames and analysis (pandas), N-D arrays (NumPy), 2D graph plotting (Matplotlib), scientific
analysis (SciPy), etc. In addition to matplotlib, other popular python libraries include ggplot (any R fans
in the group), plotly, seaborn, pygal, bokeh, geoplotlib, etc. These packages differ by their customization,
expected data input structures, charts available, export formats (e.g., svg), interactivity, dependencies
(web integrated), etc. See:
https://www.python.org/about/gettingstarted
https://docs.python.org/3/tutorial
https://scipy.org
https://matplotlib.org

Tableau and PowerBI are leading software tools for visual analytics and rapidly generate interactive
visualizations. You may download Tableau Desktop from tableau.com with a 14-day free trial, or perhaps
Tableau Cloud. Students (worldwide) get free renewable year-long licenses for Tableau Desktop (with a
valid university email address), see https://www.tableau.com/academic/students. Getting the license key
will likely only take a few seconds. If you require brief tutorials on how to use the GUI and its
functionality, see the APPENDIX A below, or visit https://www.tableau.com/learn/training. Do NOT
post and make your work visible for others to see. Tableau connects with a variety of underlying dataset
formats: offline data file on your hard drive, online databases, an online data server, or the default practice
datasets (e.g., try out the World Indicators dataset). Note how the columns from the dataset are available
on the left pane, you can drag columns to the middle pane to set the x-position y-position color size
(which automatically updates the chart graphic), dragging columns to the filter pane generates dynamic
filters within the visualization, and the right pane allows you to change the chart type depending on the
columns already selected. Users explore their dataset by creating a sheet and pairing various combinations
of dimensions and measures. For each combination of columns that you explore, switch the visualization
to several options (including Tableau's recommended chart choice which is highlighted with a red box on
the list of possible charts on the right pane). You might use this contest as an opportunity to become
familiar with Tableau, how to connect to a dataset, how to create charts (sheets), and how to create
interactive dashboards. Some of the work for each task can be done in Excel.

Contest - Rubric
The following weighted rubric details how the results will be evaluated. The final submission should be
organized in a single Word or PDF file (with legible screenshots if needed). Provide solutions that
include all the required deliverables for each task that you are able to complete. Partial solutions will be
given up to half credit. Sort your solved tasks A-G in their proper order. Clearly indicate which tasks you
attempted and which tasks you skipped. The earliest timestamp for the final pdf answer submission will
be used to break any ties.

Task A through Task G (equally weighted for each task solved correctly)

Contest – Case Study Background

Bonaire is an island in the Caribbean Netherlands (next to Aruba and Curacao) located a few miles off the
coast of Venezula. It is well-known for its healthy coral reefs, salt industry, natural park, and tourism
based on SCUBA diving and water sports. There are over 100 coral reefs locationed around the coast of
Bonaire. Let’s use visually explore marine biology data from this island (provided by www.reef.org).
Over the past several decades, the REEF organization collected underwater surveys from SCUBA divers
at each of these locations. The surveyors were marine biology experts as well as novice citizens scientists.

In the dataset we will use, Name refers to a geographic area of coral reef that was surveyed. The
corresponding Code is a unique number that was generated for each reef location. Recall from your basic
biology courses that the Taxa for family and genus are broken down into individual species. On these
survey reports, SA columns indicate that the surveyor reported the Species and the species’ Abundance
level, while SO columns indicate that the surveyor reported the Species Only and not the abundance level
of the species. The sighting frequency (SF%) refers to the percent of reports in which surveyors reported
that species. E.g., experts report seeing the Blue Tang species on 97.9% of surveys. DEN refers to the
density (abundance) of the species when it is observed. E.g., 1 means the species was solitary and only
one was observed, while 4 means a very large school of this fish species was observed.

Given the Overall_bonaire_surveys_1993_2020.xlsx file, solve as many of the following tasks as you
can. Create visualizations (e.g., a sheet within Tableau) that solve the following tasks. Turn in a Word or
PDF file with a written answer to each task and a screenshot of your visualization that visually supports
your answer. The optional Appendix B provides a very helpful walkthrough tutorial for using Tableau to
load the data, understand the case study, and answer these types of questions. If you are new to Tableau, I
highly recommend you begin by reading the Context Tasks section, and then proceed through Appendix
B, which will prepare you to solve the contest’s actual tasks.

Contest – Tasks to solve

Task A: Novice and expert surveyors are able to identify different sets of fish, which results in differing
reports. What is the top species that novices over-reported most frequently compared to experts? I.e., in
comparing the SF% for species that are more often reported by novices, which species has the largest
difference between the two groups?

Task B: Which species is ranked last with the highest Avg(Rank) (i.e., least likely to actually be identified
and reported)?

Task C: What are the top most likely family to be observed - with the lowest Avg(Rank)?

Task D: One hotel (cleverly) paid a group of expert marine biologists to survey the reef closest to their
hotel. That reef is now considered one of the most biodiverse reefs in the Caribbean due to scientists’
expertise and the hotel’s ploy. What reef name had the largest number of total expert surveys completed
(Expert SA and Expert SO surveys)? How many total expert surveys were completed?

Task E: What reef name had more novice surveys completed at that location compared to expert surveys?
How many more novices completed surveys at that location than experts?

Task F: What is the northernmost location (i.e., name of the location)?

Task G: Based on this dataset and your general background knowledge on island life, why might the
Southwest coast of the island be heavily surveyed and not the East coast? Could this bias the type and
abundance of species within the surveys of this trusted dataset?
Appendix A: Getting Started with the Tableau Desktop Interface - A Quick Guide

Select a .csv Load one spreadsheet tab from a data source at a

or Excel time. Dragging two tabs will force Tableau to
dataset perform an inner-join to integrate the data.

Pick one of your

loaded datasets to
use for this chart
1) add a second tab from the file to
integrate (join) the data or 2) switch to
sheet 1 to make your first chart
View your loaded
Change a column's datasets
metadata if needed
Make a new sheet
(i.e., chart)

New dashboard (2+

side-by-side charts)
Drag attributes
to the filter to
hide rows

Click and drag Choose the type of

attributes to either chart you want to
Columns or Rows to generate. Based on
add them to the chart your data, Tableau
will recommend a
chart type with the
red outline in the
Drag attributes to "Show Me" window.
color, size, or label
the bars, lines, and
points in the chart.
 Appendix B: Optional Walkthrough for Solving These Tasks in Tableau

NOTE: This is a teaching guide to increase your familiarity with Tableau. Don’t turn in the results from
Appendix B! Your actual tasks to complete for the contest are in the Contest – Tasks to solve section.

Bonaire is an island in the Caribbean Netherlands (next to Aruba and Curacao) located a few miles off the
coast of Venezula. It is well-known for its healthy coral reefs, salt industry, natural park, and tourism based
on SCUBA diving and water sports. Let’s use Tableau to explore Bonaire’s Marine Biology data (from
www.reef.org).

1. Download the Overall_bonaire_surveys_1993_2020.xlsx file to your computer (email

leidijon@gvsu.edu to receive a copy of the dataset if you do not have access to the files). You can
open the file in Excel to check the data.
2. Import the file in Tableau - a collection of fish surveys reported by experts and novices. Use the
Connect panel on the left to connect to a Microsoft Excel file, find and open your downloaded file:
Overall_bonaire_surveys_1993_2020.xlsx. There are three sheets of data for us to import.

On Tableau Desktop On Tableau Cloud

3. We will start by adding two sheets of data, one at a time. First, select the Species Counts Excel
worksheet and drag it to the pane on the right. Then click on Sheet 1 on the bottom menu. A sheet is
the Tableau word for a single chart. Under the Data menu at the top of your screen, select a New Data
Source, find your excel file a second time, and then add the other Survey Counts worksheet by
dragging it to the right pane. Go back to the Sheet 1 chart. After this step, both sheets (Species Counts
and Survey Counts) should be visible in the top left Data/Analytics menu. Ignore the Locations
datasheet from the Excel file for now. The data can always be reviewed in the Data Source tab on the
bottom-left corner. Every Tableau chart is designed in a separate new Sheet.
 Data/analytics menu

Data Source tab

New sheets/charts

On Tableau Cloud, the process is the same, the top menu simply looks slightly different.

Answer these questions to yourself but DO NOT turn in:

a. How many rows and columns does the Species Counts Data Source contain? Hint: use the
Data Source tab.
b. How many rows and columns does the Survey Counts Data Source contain? Hint: use the Data
Source tab.
When importing the Survey Counts sheet from Excel, note that the variable Bottomtime is a number of
hours and minutes (which Tableau sometimes mistakes for a date). In the data source pane, let’s
switch that back to Number(decimal).

c. For more details on the data source, checkout:

https://www.reef.org/db/reports/geo?end_date=2020-02-
10&format_type=chart&group_type=species&language=common&region_code=TWA&start_dat
e=1993-01-01&zone_map=0&zones=8503

4. Use Sheet 1. The columns from the dataset are available on the left pane, you can drag columns to the
middle pane to set the x-position/y-position/color/size (which automatically updates the interactive
graphic), dragging columns to the filter pane generates dynamic filters within the visualization (allowing
you to filter out undesired data), and the right pane allows you to change the chart type depending on
the columns already selected.
5. First, explore your dataset by creating a sheet and pairing various combinations of dimensions and
measures. For each combination of columns that you explore, switch the visualization to several
options (including Tableau's recommended visualization which is highlighted by a red square on the list
of possible charts on the right pane depending on the columns you selected).
Next, use Tableau to explore the dataset. I recommend making a new visualization for each task, that way you
can go back to earlier steps.

First, let’s explore the Species Counts data. In sheet 1, select the Species Counts dataset you have imported.

6. Compare novice sighting frequency to expert sighting frequency for each species. Find 5-10 species
that experts seem to identify more frequently than novices. Hint: try a side-by-side bar chart and show
individual datapoints (click Analysis in the top menu and then unselect Aggregate Measures).
d. You would ideally represent the insight in the simplest visualization that gets the story (key
takeaway) across. Write a short title that clarifies what the key takeaway is for the reader.
Remember to add titles to your work, e.g., “Differential fish counts by expertise level - Leidig”.

Add your lastname to

your charts for full credit

7. Using the new chart, how much more do experts seem to identify these species more frequently than
novices? A calculated field would make this much easier to solve. Right click in the left pane within the
Tables section and select Add Calculated Field… Write a simple calculation to generate a new column
(DifferenceInSF) for each row in the data table that subtracts Novice SF% from Expert SF%, calculating
how much better experts are at sighting each species than a novice. It looks like experts identify and
report Rainbow Wrasse more often than novices.
8. Create a new sheet (Sheet 2). I recommend making new sheet and chart for each question. Some
species are very prevalent and are observed on almost every fish survey. Which species are seen on at
least 90% (or more) of total surveys? Show all of the ~11 species. Hint: create a chart for species and
TotalSF%, drag the field Total SF% to the Filters pane, and then only show species found at least 90%
of the time.

Sort the data

9. Recall from Biology, multiple species are organized into a family of similar organisms (Homo sapiens
fall under the Hominidae family). E.g., Brown Chromis and Bicolor Damselfish are both grouped into the
Damselfish family. The species with a rank of 1 is seen most frequently and the species with a rank of
547 is seen the least frequently.
e. Generate a new sheet/chart that shows the average (avg) rank of the high-level family groups.
Hint: you can sort the data within your charts to make it easier to answer the questions.

f. Hint: generate a graph of family compared to Avg(Rank). Tableau defaults to calculating the
Measure (Sum) instead of Avg – but that can be changed in the dropdown from SUM(Rank).
g. What are the top 5 most likely families to be observed - with the lowest Avg(Rank)?

10. Dragging the Species dimension on top of the Family dimension in the left pane will create a hierarchy.
Thus, the Family category will be the parent to the Species sub-category (Family,Species).
 h. Generate a new Treemap sheet showing the average total sighting frequency and average total
density for each family,species using size & color. It may be helpful to set SF to area and DEN
to color.

i. Which families have the highest densities? A schooling family like silvery bait fishes perhaps?

11. From the same style chart, click family to expand the category to show the breakdown for its species.
 j. Find the Trumpetfish sub-category – it may take some hunting! Let’s compare the Trumpetfish
family to the Damselfish family. There is only one species of Trumpetfish in our data while there
are 14 Damselfish species. Pretty neat!

Next, let’s explore the Survey Counts dataset. Create a new sheet and select the Survey Counts dataset.

12. SA columns indicate that the surveyor reported the Species and their Abundance, while SO columns
indicate that the surveyor Only reported the Species they saw and not the abundance level. Name
refers to the name of a coral reef that was surveyed. Code is a unique number for each reef site.
k. Let’s figure out whether experts or novices are providing data for each reef. Make a side-by-side
bar plot that compares the total (Sum) number of Expert SA reports to the total (Sum) number of
Novice SA reports for each reef site.

l. Let’s assume that experts provide more accurate survey data than novices. Are there any reef
sites that appear to have significantly more data from novice surveys than expert surveys? If so,
analysis regarding these sites may be incomplete or misleading! List any such reef names (hint:
~3-5 locations).
m. Side note: On average, experts report 80.86 species per survey and novices report 55.64
species – this could bias the overall data.
13. Let’s create a calculated field that counts how many total surveys are available per location, regardless
of the person. To do this, we will add a new measure that simply adds the ExpertSA, ExpertSO,
NoviceSA, and Novice SO columns.
n. From the top menu, select Analysis >> Create calculated field, and add the equation.
 o. Rename Calculation1 to TotalSurveys.
p. Create a sheet of the total surveys per reef name using a chart type of your choice, e.g., bar.
q. Sort the reefs by deceasing numbers of TotalSurveys in the top menu. This menu option can

sort ascending, descending, or transpose (flip the rows and columns).

r. What are the names of the top seven reefs by survey count? Were any of these locations
identified as being possibly biased due to too many novices from question 10?

Next, let’s place these datapoints on a geospatial map.

14. Add a new datasource out of the Excel file’s ‘Locations’ worksheet. Open the Excel file as a new data
source. Drag both Survey Counts and Locations worksheets into the top pane.
15. Rows from two separate datasources can be merged into a larger row as long as they both contain an
identical identifier that can be matched. We are going to join the survey data for each reef location (a
row in the Survey Counts table) with its actual Lat/Lon coordinates (from the Locations table). Edit the
relationship to force Tableau to only match rows from the Survey Counts and Location datasets
together if they discuss the same location Code. Each reef site gets its own unique code and name. In
the database world, this is called in inner join, natural join, and equijoin.

16. Edit both Lat Dec and Lon Dec columns so that Tableau interprets these as decimal numbers and as
latitude and longitude data points. Also, select to use them as columns with actual Lat/Lon coordinates.
You also may have to convert them between discrete/continuous/measure/dimension.

On Tableau Cloud, the menu may be very slightly different. You must make both fields decimal
numbers and also assign them a latitude or longitude geographic role.
17. Create a new sheet for our map from our new Survey Counts+ dataset. Drag Lat Dec and Lon Dec to
the chart. You may have to edit the two pills to convert them both to Discrete if the map is not
correct. Tableau generates an error, reaggregates the data, and then works fine if you try again.

18. If Tableau interprets your lat/lon coordinates correctly, it will generate a map with a point for each row.

19. Add Expert SA to set the size of each point and Name to each point’s label. Hint: On the left Tables
pane, you may have to select the Lat Dec and Lon Dec dropdown menus and Convert them to
Discrete. When you add Expert SA to the Marks pane, you may have to change Measure(Sum) to
Attribute.
20. Where do experts seem to conduct the most surveys?
21. Reflect+answer: Why might the Southwest coast of the island be heavily surveyed and not the East
coast? Could this bias the type and abundance of species within the surveys of this trusted dataset?