Creating Data Visualizations of The Bachelorette
— Introduction
As part of our HCI curriculum, we undertook the “Design for Understanding” project. This project was centered around mapping data to visualizations in order to practice communicating and presenting data.
Our group chose to analyze information on the TV shows The Bachelor and The Bachelorette. The concept of these two shows is that a single contestant dates a pool of approximately 25 people at once. Each week the bachelor/bachelorette will go on dates with the contestants and decide which ones they are still interested in. The lucky contestants are kept on the show via a rose given to them from the bachelor/bachelorette. This continues until the final week when a winner is chosen and a proposal is made.
The data itself was focused on each contestant of either show, the types of dates they went on while in the show, and the result of each contestant each week (either being eliminated or winning). We chose this data set because the two shows have a lot of fascinating trends that would allow for interesting stories to be told through the data. The show is watched widely every season it comes on with people tuning in week by week to see the results. Watchers develop personal connections with the contestants, pick favorites, and hope that these contestants will make it to the final rose ceremony. Our hopes by picking this data set was to create graphs that would showcase some of these trends, and perhaps even inform people enough that they can make more educated predictions for future seasons.
We produced two types of graphs, each communicating the same data for different purposes:
- For Clear Communication: A series of graphs that tell the story of the data. A minimum of 3 charts that explore the data from different perspectives.
- For Persuasion: Few guidelines, made to be creative and impactive.
For our communicative graphs, the final product consisted of several graphs that tackled different parts of the data, the links to view these can be found here:
The communicative graphs are:
- Season Events: This chart plotted the combination of all events that happened within a given week of any season. An “event” was defined as either a rose being granted, or an elimination. These events were paired with the types of dates that a contestant went on.
- Eliminations Per Week: A bar chart that breaks down the number and type of eliminations per week. The bars represent the sum of eliminations for a particular week.
- Elimination Comparison: A bar chart that compared the types of eliminations between the two shows, season by season.
- Roses Given: Technically two charts, one a bar graph and another a line graph. These charts depict how roses were distributed for each show.
- Dates of Winners: A bar chart that shows the types of dates the winners of each season went on.
Our Persuasive graph: The Contestants of The Bachelorette focused a lot more on the visuals, taking advantage of the show’s love imagery, particularly with roses and hearts. The chart depicts every contestant of both shows and marks their progress through the weeks. Depending on what happened to a particular contestant on a particular week, there are icons depicting awarded roses, eliminations, firings etc. leaving only the victor at the end. The purpose was to show a visual representation of the journey each contestant goes on throughout the weeks of the show.
— Preparation
We decided that our graphs would be produced in Tableau since we had an in class demo on it. The various features such as tool-tips and filters that could be added to the data were perfect for the types of graphs we wanted to produce. The goal was to produce some simple to read graphs, that would show much more detailed information in the tool-tip. We wanted this data to be filterable so that viewers could search through various weeks, seasons or even contestants. Our group had to get appropriate data to produce the graphs, though.
To start, we gathered data on the show from Datasets underlying FiveThirtyEight stories provided on the class webpage and attempted to export this data directly into Tableau. We started producing simple graphs with this data until we noticed some of the data was missing. In previous seasons, eliminations were being switched to null data points. This would make most of our data incomplete. To fix the error, we had to export the data into excel and then use the excel data in Tableau. This step allowed for all of the data to be translated into Tableau.
— Brainstorming and Prototyping
Our design process was based on the five design-sheet for data visualization. This process consists of five main stages:
- Meet
- Brainstorm
- Design-sheets
- Discuss with client
- Generate Realization Design
The process was further subdivided at the brainstorm and design-sheet steps, however we did not follow each of those substeps discretely, preferring instead to tackle issues that we found throughout our discussions.
During our first meeting we brainstormed what types of graphs we wanted to produce and the types of information these would show. Initial suggestions were for graphs showing total eliminations, graphs showing the dates that winners went on, and graphs with clips of contestants as they were kicked off of the show. We also decided it would be interesting to make a graph with various filters to find specific information about the show. We wanted to include filters about seasons, which show, and contestant names.
As we were having these discussions however, we were also looking at the data itself, and the way it was formatted, for the first time. This meant that many of our ideas were vague and light on the actual technicalities of how they would appear. This was to the benefit of the brainstorming process, which is meant to inspire quantity rather than quality. However it also meant that, later on when actually being more familiar with the data and how it is represented within Tableau, we were forced to make modifications or scrap entire ideas altogether.
— For Clear Communication
We split the team into two groups to focus on our communicative design and our persuasive design. Dan and David worked on graphs for the communicative design. These would be all of the graphs that were purely based on data. The data we had were contestant names, eliminations that happened each week, and dates the contestants went on each week. The way a contestant was eliminated appeared in several forms:
- E - Contestant was eliminated at a rose ceremony.
- ED- Contestant was eliminated at a date.
- EU - Contestant was eliminated at an unscheduled time outside of the show’s episodes.
- EF - Contestant was fired by production.
- EQ- Contestant quit the show.
The information about dates simply included how many people went on each date, such as if it was a one on one or a group date with eight people.
This type of data led us to create graphs showing eliminations and roses that occurred throughout the show, as well as the types of dates the contestants went on. We were also able to filter out winners of the show to give a specific graph of the dates they went on. We chose to use a bar graph for many of our graphs to easily convey differences to the viewer. We learned in class that humans are very good at telling the relative difference in length between two bars, making bar graphs optimal for comparing different quantities.
To depict the entire data-set as an image we developed the Season Events chart.
Every marker on the graph encodes a wealth of information. First, the marker itself depicts which show the event occurred in; a square is used for The Bachelor and an x is used for The Bachelorette. The type of event (elimination, receiving a rose, etc) is simultaneously encoded by the color of the marker as well as which subsection of the chart it occurs in. The height of the marker within the subsection also encodes the size of a date (if any) that occurred along with it. Lastly, the horizontal position of the marker shows in which week the event happened.
The chart is also equipped with several ways to filter the information to areas of interest. Specifically, a user can change what shows, seasons, and contestants are displayed. The chart is a complete visual source of information for the bachelor and bachelorette, but isn’t particularly good at showing trends in the shows, so other graphs were developed to provide better insight into the data.
To show specific trends within the data, we created several different graphs. Some of the graphs focused on eliminations, such as the graph showing eliminations per week.
This graph shows the obvious trend of declining quantities of eliminations as the show progresses, mainly due to the fact that there are fewer contestants remaining each week.
Other graphs include the roses given and roses given over time, which focus on the roses given out by the bachelor or bachelorette, rather than on eliminations, or the graph depicting the types of dates the winners went on throughout their season. This graph was very enlightening, showing that very few contestants who have ever gone on a 2 on 1 date went on to win the show, and in fact no one has ever won The Bachelorette who went on a 2 on 1 date during the season.
— For Persuasion
Daniel and James did work on the persuasive graph. The initial idea for this graph showed up in one of our brainstorming sheets, it was supposed to have images of all of the contestants as well as clips of the eliminations, along with icons for depicting a particular contestant’s elimination.
However this was scrapped because the faces and clips would have to be found online and added individually for every contestant, this would have been an extremely even without Tableau, as there have been hundreds of contestants throughout the years. Not to mention that Tableau itself is very finicky with inserting custom images for data points.While it would have added to the emotional connections of the show, it turned out to be too much work for the short time span of the project.
The graph was made to show whether a person was eliminated or got a rose every week. The goal of this was to highlight the emotions felt by the contestants going through every week of the show. Through our reading and class lectures, we found that graphs with interesting imagery (such as dinosaurs) leave a more lasting impression on people who viewed them. We wanted viewers to develop more of a connection with each contestant than they did with the purely informative graphs. We chose to convey this message with images associated with the heartbreak on the show. These symbols would be more eye catching similar to the monster graph and the graph following the curve of the woman that were shown in class.
We chose the icons specifically based off of the types of eliminations that occur on the show. The roses were represented with a rose icon to create a visual of the ceremony on the show. Regular eliminations were represented with a broken heart icon. This was chosen deliberately so that viewers would relate heartbreak to the elimination each person went through. Contestants that were eliminated during a date were represented with a rejected symbol. This was meant to show the harsh interaction between the bachelor/bachelorette when they have to kick someone off during a date. Occasionally, contestants have been fired from the show. This was shown with a fire icon to create the relatable feeling of someone being fired. The last way people are kicked off is from mysterious circumstances. This was represented with a large question mark showing the confusion associated with this contestant leaving the show. Finally, the winner of the show received an icon of engagement rings to show the proposal happening on the final episode.
The tooltip was also an important focus of this graph. A user looking through might not know what each of these icons represents, but through Tableau’s tooltip feature, readers could hover over any icon and receive more information as to what that icon means. These were used to clarify information for viewers that were unfamiliar with this TV show.
— Feedback and Reflections on our Presentation
Some of the main concerns from our demo included people being worried about too much clutter, wanting the names to be organized better, wanting better label names, wanting better use of the tool tip, and wanting an inclusion of images and clips of the contestants. Some of the feedback was used to change our project while some was purposefully ignored.
The clutter was seen as a serious issue. For viewers unfamiliar with the show, the graphs could become overwhelming. To combat this issue, some of the main graphs were split into multiple charts to make the data easier to understand. We also agreed with the issue of the names.
The names were originally in a form that showed the season and contestant name all as one word with underscores for spaces. While it conveyed a lot of information, it was visually unappealing, and it distracted from the overall graphs. A code was then implemented to change all of the names to only show the first name and last initial of each contestant. This was used throughout all of our various charts and graphs to make them more uniform. Chart and graph labels were also modified. Originally, they were depicted as their encodings within the data-set, meaning labels such as won, eliminated, and fired would show up as W, E. and EF respectively. This made the graphs impossible to read without an explanation from one of us. After receiving user feedback, the labels were corrected to better represent the event they encoded.
For the persuasive chart we agreed that adding in pictures and clips of contestants would be helpful and add to the memorability of the graph, but this task turned out to be too difficult for the scope of the project. None of these images or clips was provided in the data, so it would have to be found outside of the data. Many clips from the show were unavailable to save or not even available online anymore. The clips that were found wouldn’t play in Tableau. This caused the idea for clips to be dropped. The images idea went through the same sort of issues. These various issues made the idea for pictures an impossibility as well.
— Conclusion
Working with data-sets and creating visualizations was something that was brand new to most of our group, ensuring many challenges and hiccups along the process. Working with Tableau, a program unfamiliar to all of us even with a basic tutorial session, was the hardest challenge to overcome. Translating all of our ideas into decent looking Tableau graphs took a lot of time. The persuasive graph in particular had a lot of customizable aspects that were not covered in the tutorial, and so required a lot of external research to implement correctly.
Overall we were pleased with how we handled the data, and believe that we did the different types of representations justice based on what we were working with. The Bachelorette and The Bachelor made for fun graphs with interesting trends and stories to tell.
___________________________________________________________________
Tableau Links:
Team Members: Daniel Perez, Dan Kershner, James Capone, David Schwartz
