Week 12: In-Progress Crits

Introduction to Data Visualization

W4995.003 Spring 2024

​

​

Guest Critics

​

THIS WEEK

• Matthew Conlen
• Data journalist (Our World in Data, New York Times, 538) and computer scientist (Jeffrey Heer’s lab)

​

NEXT WEEK…

• Asad Pervaiz
• Creative director and graphic design professor at M/I/C/A & Pratt Institute

​

IN TWO WEEKS…

• Klara Auerbach
• Illustrator and data visualization designer (Our World in Data, The Atlantic)

​

​

Time Limits

​

• 5 min. Presentation

​

• 7 min. Feedback

​

​

Use the feedback form for any comments you weren’t able to make in class.

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

Mapping Campus Safety

Table of contents

01

02

03

04

05

Motivation

Prior Work

Exploratory Visualizations

Sketches

Questions

Motivation

Safety Concerns

• Traveling at night
• Areas to avoid
• Route selection

Goal

• Spread knowledge of high risk travel areas and times
• Safer pedestrian travel

Data Story

​

We are informing pedestrian travel at Columbia University to enable safer student transit.

Wireframes

​

Wireframe

D3 Visualization

​

Questions

1. What do you think is the most engaging way to gamify a map?
2. How would you design the user interactivity?
3. Is there a way to make location data more attractive?

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

5.2: In-Progress critique

nna2132, rgd2127, na3062, tb3061

​

Motivation

To provide a comprehensive view of crime trends in the US over four decades, presenting data in an accessible manner to a wide audience. This could help demystify perceptions about crime rates and their fluctuations over time.

Hypotheses

Trend Over Time: Fluctuations in crime rates over four decades reflect underlying historical, economic, or policy changes.

Geographic Variance: Significant differences in crime rates are evident across regions, with certain areas consistently showing higher or lower rates.

Population Impact: A correlation exists between population size and crime rates per capita, suggesting different crime dynamics in larger vs. smaller jurisdictions.

Type of Crime Evolution: The prevalence of specific violent crimes (e.g., homicides, rapes) has changed over time, indicating shifts in crime patterns.

Relevant prior work

FBI's Crime Data Explorer Inspirational for its comprehensive approach to presenting crime data.

https://cde.ucr.cjis.gov/LATEST/webapp/#/pages/home

Relevant prior work

"The Next to Die" by The Marshall Project�An example of how to effectively use narrative and visualization for exploration.

https://www.themarshallproject.org/next-to-die

Relevant prior work

Out of Sight, out of Mind�Inspirational for its animation narration

https://drones.pitchinteractive.com/

Exploratory visualization

Hypothesis 1: Trend Over Time – Fluctuations in crime rates over four decades reflect underlying historical, economic, or policy changes.

A line chart showing the overall crime rates in the US from 1975 to 2015.

​

Exploratory visualization

Hypothesis 2: Geographic Variance – Significant differences in crime rates are evident across regions, with certain areas consistently showing higher or lower rates.

A bar chart showing the overall crime rates in different states of the US from 1975 to 2015.

​

Exploratory visualization

Hypothesis 3: Population Impact – A correlation exists between population size and crime rates per capita, suggesting different crime dynamics in larger vs. smaller jurisdictions.

Stacked bar chart from 1975 to 2015 reveals correlation between population size and crime rate per capita which changes over time for each state.

​

Key Takeaway from our visualization

Over four decades, the landscape of US crime reveals a nuanced narrative

1. Explore Yearly Trends: A mid-90s peak in robberies and assaults reflecting societal shifts
2. Analyze State-by-State Data: A compelling correlation between population size and crime rates, underscoring the diverse crime dynamics across states and counties.

Data Story

An intricate tapestry of societal change, seen through the lens of crime data, underscores the evolving battle against crime in the U.S

Rough Visualization in Code

Sketches of how final visualization will interact

Feedback on…

Q1. Is our narrative cohesive? And how can we make it better?

Q2. What common pitfalls should we avoid in data storytelling?

Q3. Could you please advise on whether it's better to present information regarding numerous counties across the country at once, or would it be preferable to structure the data by state, offering detailed graphs for each county upon the selection of a state?

Q4. When aiming to cater to an audience without expertise, which type of visualization would be most easily understood: a choropleth map, a cartogram, or a bubble map?

Thank You

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

Visual Insights from a Decade of Citi Bikes in New York

Joric Barber, Franco Magalhaes, Martha Njuguna, Margot Stern

1: Initial Hypotheses

1.1 Motivation

We want to highlight how Citi Bike use has evolved in the past 10 years, offering the City an effective transportation alternative that serves different New Yorkers in different ways.

There are also ways in which the service falls short: Citi Bike is worse in low-income neighborhoods, not offered in some neighborhoods at all, and undermined by the City’s failure to build protected bike lanes.

1.2 Hypotheses

1. Across Time: Investigating how Citi Bike usage evolved over time, including the hypothesis that ridership spiked post-COVID lockdown in 2020, as well as exploring correlations between ridership and changes in public transportation costs.

​

• General Statistics: Exploring areas of NYC with varying likelihoods of Citi Bike usage and their interaction with social factors, analyzing usage patterns throughout the day, identifying peak bike traffic times and areas, and examining the demographics of the most common Citi Bike users across gender and age.

2: Relevant Prior Work

2.1 Reference

In this New York Times article, interaction design is paired with the text in a way that amplifies the reader’s understanding of the information relayed.

2.2 Reference

In this interactive exploration of Boston’s subway system, the scroll/hover feature facilitates an in-depth analysis of transportation data throughout time.

​

2.3 Reference

In this explanatory visualization of delivery workers’ working conditions, mapping and scrolling features are jointly employed to relay routes and various figures related to overtime, instances of theft, and work-related expenses.

​

​

3: Exploratory Visualizations

3.1

Leading up to the COVID-19 pandemic, men made up the bulk of Citi Bike’s user base. Ridership levels peaked in 2018 before steadily decreasing until 2020.

3.2

From 2013 until 2020, a large majority of rides departed from stations in lower- and mid-Manhattan.

3.3

Rides tend to peak at 8AM and 5PM daily.

4: Data Story

Since its launch in 2013, Citi Bike has amassed a diverse ridership and fundamentally transformed how people navigate and experience the largest city in the United States.

5: An Initial Viz

6: Sketches

7: Feedback

Questions

1. Since there are so many dimensions to the data (the types and gender of riders, types, durations and locations of trips), which dimensions do you think are most important to focus on?
2. What are potential limitations in the data or visualization that we should be aware of, and how could they be addressed?
3. How do we balance/accomplish relaying the more pragmatic, high-level insights from the data alongside a more “fun,” personable view of how Citi Bikes play a role in people’s day-to-day lives?

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

Nutritional Profiles of Starbucks Drinks

Ha Yeon Kim, Claire Chen, Anusha Lavanuru

Data Story

Jennie, a Starbucks enthusiast, is currently on a diet wishes to check the nutritional information prior to ordering her beverage.

​

Given that she typically customizes her drink, relying solely on the app is insufficient for her needs.

Motivation & Hypothesis

Problem

The current Starbucks menu only shows the total calorie per drink, helpful for controlling calorie intake but lacking depth in terms of comprehensive nutritional facts crucial for maintaining a healthy lifestyle.

Moreover, the standard menu presentations fail to account for variations attributable to customizable options, such as differing milk choices, presenting a significant information gap for people seeking to make informed dietary choices.

Solution

In response to this gap, we aim to build a tool for visualizing the nutritional facts of Starbucks drinks to enhance consumer awareness and empower people with the knowledge to make healthier choices.

Hypothesis

1. There are notable differences in the nutritional content across different categories of drinks.

​

• Substituting traditional dairy milk with alternative milk options and variations in sugar levels significantly alters the nutritional profile of a drink, particularly in terms of calorie, fat, and carbohydrate content.

​

• Seasonal drinks offer unique nutritional profiles compared to standard menu offerings due to the incorporation of special ingredients.

Exploratory Viz/

About our dataset

Prior Relevant Work

https://rpubs.com/saravargha/starbucks

https://mavenanalytics.io/project/730

​

Final Viz Prototypes

Draft Viz (+code)

https://observablehq.com/d/9b0c47f024e5a8ee

​

https://observablehq.com/d/1561998a18464c05

Questions

Questions

1. Any suggestions on how to visualize the decision-making process when ordering a drink?
1. dropdown menus
2. select in the tree structure
3. drag and drop
2. Would you recommend the food menu as well so that users can form combos and check out the total nutritional profile?
3. With the dataset, is there any other data story that can be used?

Thank you

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

A5.2 Project Proposal

Natalie Bran, Rosa Figueroa, Yueqi Li, Lindsey Cruz Rosales

Motivation

"March Madness showcases the pinnacle of collegiate basketball talent and dedication. Our project aims to explore and highlight this tournament's intricacies, inviting a broader audience to share in the excitement and celebrate the hard work of its teams."

Hypotheses

• Certain conferences consistently demonstrate higher win rates in March Madness.
• The average seed rankings within certain conferences will be significantly higher, suggesting these conferences are perceived to have stronger teams overall.
• There is a strong positive correlation between higher seed rankings and a team's win rate and likelihood of advancing in the tournament, indicating that seedings are a reliable predictor of tournament success.
• Certain seeds are more prone to being involved in upsets either as the unexpected victor or as the higher seed that gets defeated, which could challenge the reliability of seeding as a predictor of success.
• Testing these hypotheses with the data can lead to insightful conclusions about the nature of competition in March Madness, the effectiveness of the seeding process, and the dynamics of success within and across conferences.

Relevant Prior Work

​

​

Display of winning schools over the years including their seeds and number of times won

​

​

Visualizes the Men's and Women's NCAA Tournaments as radial brackets

​

​

Relevant Prior Work (Cont.)

​

​

Adam Pearce’s work displays how NBA playoff probabilities shifted since the beginning of the 2019-2020 tournament

​

​

​

​

The Pudding’s essay on how artists get paid from streamed – Inspiration for scrollable, animated essay style

​

​

​

​

Exploratory Visualization

Number of championship wins of each conference

The average ranked seed for teams in each conference

Conference Results

Exploratory Visualization

Wins and Performance Against Seed Expectations by Seed

Total non-zero wins by team by seed

Upset and Seed Analysis

Exploratory Visualization

Closer look at seed 11 performance by teams ranked seed no. 11

Frequency of upsets for certain winning seeds and seed matchups

Number of Upsets from 2008-2023 by Unique Seed Match-Ups

Number of Upsets from 2008-2023 by Winning Seed

Upset and Seed Analysis

Rough D3 Graph

Breakdown of championship conferences and teams (2008-2023)

Rough D3 Graphs

Interactive visualization that shows 20 years of basketball winners

Bubble chart linked to line graph that shows the relationship between conferences, teams, and seeds in the tournament from 2008-2023

Rough D3 Graph

Search function across 2008-2023 championship data

Rough D3 Graph

“March Madness thrives on its thrilling unpredictability and historic upsets, and our data story reveals how some conferences consistently outshine the rest, while pinpointing the most astonishing upsets ever witnessed in tournament history.”

​

Our Data Story:

Sketches

Sketch

Additional Idea!

Feedback Questions

• In regards to March Madness, what aspects specifically spark your curiosity?
• Is there any area you think could be interesting to explore?
• Did our story flow logically for you, and did it make sense in its progression?
• Is there any aspect of our narrative you think could be improved or altered for better understanding?
• Do you envision alternative ways to represent this data that could enhance understanding or engagement?
• Do you have any suggestions or ideas for interactive elements you'd like to see?

​

​

​

Thank You!

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

A5 In-Progress

Geneva Ng, Aparna Rajesh, Sam Miserendino

Motivation and Hypothesis

We are concerned about the safety of public transportation in New York City and hypothesize that there have been elevated levels of violent crime in the subway in the city over the last 6 months.

We're investigating whether:

• this rise marks a record high or a recent peak
• such crimes are increasing in isolation
• there's a general uptick in violent crimes citywide during this period

Prior Work

Anton Bardera’s Multidimensional Visualization: Linked Views will inform the setup of our linked view, especially in connecting two views, with Chazan’s work further guiding the presentation of linked data.

In the center, @kiko-datasparq’s Pricing Algorithm visualization will guide our main visualization, offering an easy comparison of two data sets on the same axes.

Adam Chazan’s MD Countries Total Cases Map will inspire our linked view opacity map.

Exploratory Viz 1

Exploratory Viz 2

Exploratory Viz 3

Our Rough D3 Viz

Final Viz Ideas

Final Viz Ideas

Feedback Questions

1. Are any other ways that we can possibly combine or pre-compute the values in these visualizations to make the default view more comprehensive?

​

• Are there other data sources that would be better / limitations in official reporting data like this that need to be considered? It seems like the authoritative and only dataset you’d need at first, but this could be a bad assumption.

​

• (on the next slide)

Feedback Questions

Are there any dimensions to the data that these columns would make you curious about? Aka are we missing any interesting data you’d want to see?

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

A representation of art and artists at the MoMA

Olivia Schmitt, Lucia Perez Saignac, Racquel Lemoine

Motivation and Hypotheses

The representation of different genders and ethnicities within the MoMA’s collection raises important questions about diversity and inclusivity within the art world. This proposal aims to investigate the representation of artists in MoMA's collection, focusing on gender and ethnicity, and how these have changed over time.

​

• Are certain genders or ethnicities underrepresented in MoMA's collection? How has this representation evolved over time?
• How has the scope of MoMA's collection changed over time? Which periods witnessed the most significant growth or decline?

Relevant Prior Works

Prior work by FiveThirtyEight has investigated different factors related to the changing composition of the Moma’s collection over time, such as the painting size, year painted, as well as artist’s nationality and choice of medium. We want to expand on this work by exploring the intersections of these factors, specifically as they relate to the identity of the artist. Thus, an important aspect of our data visualization is leading the reader to discover new artists as they uncover their own trends and gaps within the data.

Relevant Prior Works - cont.

https://ryanabest.com/ms1-2018/quant/final/

https://fivethirtyeight.com/features/a-nerds-guide-to-the-2229-paintings-at-moma/

Exploratory Visualizations - I

Exploratory Visualizations - II

Exploratory Visualizations - III

Exploratory Visualizations - IV

Sketches for Final Viz

Sketches for Final Viz

What do we want our readers to take away?

Data Story

Readers should walk away with an awareness of the lack of diversity at the MoMA, an understanding that while there is some type of initiative to increase representation, it is still very much lacking.

Initial Visualization

Python Generated Visualization using Jupyter Notebook

Feedback: Specific Questions

Scaling

Interpretation

Engagement

Are there any common difficulties that are faced when it comes to analyzing gender and heritage that we should be aware of? What ethical considerations should we keep in mind?

Considering there is a large range in our data (male vs non male artists) it is difficult to display these data points without overpowering the smaller values. What type of visualizations would you recommend to combat this?

What strategies would you recommend to appeal to a wide range of people and not just those involved in the art world?

01

02

03

Thank You!

Please keep this slide for attribution

CREDITS: This presentation template was created by Slidesgo, and includes icons by Flaticon and infographics & images by Freepik

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

Is there a way to

engineer a top hit?

Arthi Krishna, Rachel Michaelson, Elaine Su, Abhishek Chaudhary

Why this topic?

Motivation & Impact

As avid music listeners, we were curious as to if there were reasons why certain songs were hitting the top charts and what that said about people’s music taste through the decades.

We aim to provide a comprehensive exploration of what makes a song successful,

offering valuable insights for music industry professionals and enthusiasts.

​

​

Our Dataset

Top 10,000 Songs on Spotify 1960-2023 (from Kaggle)

Songs (according to Spotify API) are evaluated by several metrics, including

Danceability

Acousticness

Instrumentalness

Loudness

Valence

Speechiness

Liveness

Energy

DataViz Explorations

Analysis Findings

DataViz Explorations Continued

Analysis Findings

Technical Visualization

Test Created in D3

Lo-Fi UX

Sketches

Hi-Fi UX Mockups

Intro + Exposition

Hi-Fi UX Mockups

Decade Timeline

Hi-Fi UX Mockups

Final Personalized Exploration

Feedback Questions

What do we need help on?

Best ways for multi-view coordination

Emphasizing patterns in the data

Overall planned user experience & interactiveness

Thank You!

Any Questions?

01 Mapping Campus Safety

02 US Crime Rates (1975–2015)

03 NYC Bike Ridership Patterns

04 Nutritional Profiles of Starbucks Drinks

05 March Madness

06 Violent Crime & the Subway

07 Representation of Art/Artists at the MoMA

08 Engineering the Perfect Track

09 The Art of Perfecting Sleep

​

​

Slumber Stats:

​

Plotting Your Path to Quality Sleep

By Emily Xia, Chengke Deng, Jyothi Gandi, Kentrie Tran

What is one experience everyone shares and craves?

The Answer: A Good Night’s Sleep

What does it mean to have a good night of sleep?

• Sleep Quality (user-measured)
• Sleep Duration
• Sleep Efficiency (amount of time spent in bed)

How can we sleep better?

• Narrative focused on YOUR sleep routine and how it compares to the data
• “Choose Your Own Adventure” style exploration

Our Inspiration

​

“The Secrets to Good Sleep” (NYT)

“Sleep Better at Every Age” (NYT)

“Relationships between sleep efficiency and lifestyle…”

(Yu Ikeda, et al.)

Datasets

Hypothesis and Questions

Sleep Factor Correlation

Interactive Visualization Draft

"Your sleep journey is unique and shaped by your habits - from phone usage to alcohol consumption, the quality and duration of your restful nights depends on YOU."

Feedback Questions

THANK YOU FOR LISTENING!