Half the teams from our 100-person class chose to feature their projects in this showcase. Here is their work!
A tidal wave of statistics, fatality rates, and haunting predictions of the future consumes our media and have the placed a cloud of uncertainty over the world. I was inspired to make a project related to COVID-19 but did not want to make a typical visual exploring numbers of cases or deaths; I wanted this to be more human and personal. Cabintine is a data physicalization made out of found materials (cardboard and cereal boxes) in the form of a 3D bar chart that allows users to physically touch, compare, and filter data. The piece explores a personal dataset: a daily log of 11 students living out COVID-19 quarantine in a remote cabin. The log details how they are doing and the events of daily life under these unprecedented circumstances. Cabintine facilitates multiple modes of physical interaction, allowing users to explore the structured and unstructured data in a way that is much more tangible and flexible than methods afforded by digital interaction.
The purpose of this project is to explore ethical visualization design guidelines in the context of exonerations in the United States since 1989. The resulting visualization includes a slideshow with interactive elements meant to serve as an introduction to the topic of exoneration and an interactive data visualization in which users can explore exoneration data from The National Registry of Exonerations and view stories from exonerees. Through filtering, grouping, and informative text, the user is able to gain an understanding of the data behind exoneration and find unique insights and trends. This visualization experiments with different encodings meant to humanize the data, and serves as a case study for what an ethically designed visualization might look like.
What characters are used most often in Chinese and how are they used? Are there specific types of characters more likely to appear? This project explores the frequency of Chinese characters and how they can combine into various words. You can examine the usage of different characters and find other characters related in meaning.
In this project, we have developed an interactive datavis game to create awareness about taking precautions against spread of COVID-19. Our COVID-19 game allows the users to understand the effect of precautionary actions taken by them and the community by simulating the daily spread on COVID-19 through a game. The users can first select different levels of precautions such as wearing masks, washing hands and social distancing, then they can simulate the spread and finally observe the impact of their actions on number of infections with a daily graph. Thus, we use a gamified narrative, interactive simulation and statistical visualization to create personal awareness of taking the precautionary actions to avoid the spread of COVID-19. To evaluate the experience of our visualization game, we conducted an informal user-study with 10 participants and collected their qualitative feedback.
Self-similarity matrix representations of music has been shown to enable identification of structural and rhythmic elements by visual inspection. However, previous efforts in visualizing symbolically encoded music this way aim to visualize all, or most, instrumental parts of the opus as a single image, by overlaying multiple similarity matrices and introducing a myriad of color encoding. The result is something that can be difficult to decipher. Furthermore, from just the standalone matrix image, it becomes difficult to pinpoint the exact locations in the original piece a particular matrix cell is meant to represent. Here, we propose an interactive music visualization system that addresses these shortcomings. We anticipate that this visualization system would enable even laymen who cannot read musical notation to track and appreciate the structure of a piece of music.
Through our visualization, we wanted to explore the common environment and features that facilitate scientific creativity and increase the academic achievements of researchers. The best examples to illustrate this is by looking at Nobel laureates to see how and in what conditions they won their awards. We visualize information about the laureate fields, the age they won the prize, the number of publications they have, whether they stayed in their born country or they moved to other countries, and whether they collaborated with other researchers or if they worked alone. We illustrate this data in the shape of a flower and each flower represents one laureate. We used scrolly telling to teach our encodings to our users step by step, and then after learning about that, they can explore our garden. We used individual graphs for each laureate to humanize the data and engage interactivity. Also, users can explore each story separately and learn about each laureate.
We present RIDE: The Animation, an interactive visualization that uses the dataset provided by the City of Chicago on trips taken using ride-hailing services from November 2018 onwards. This is the first interactive visualization that we are aware of for this dataset. Our visualization uses animations to show the start and end of individual trips as well as the speeds of trips, and features a playful design that entices users to engage with the dataset. We show that RIDE enables users to take an exploratory approach to the complex dataset and find basic trends that may inform future mobility decisions.
In this study, we design and conduct experimental study to analyze the impact of a series of different visualization methods on user's behavior change. Specifically, we focus on analyzing the effects of different screen time visualization and user's smartphone usages, to compare the persuasive power of good visualization against bad ones.
Due to Covid-19, cities around the world are shutting down and citizens are staying home. In April, the United States became the country with the highest number of confirmed cases, and New York City became the world's top Covid-19 hotspot. In our work, we analyze how New York City's subway station density has changed over the last few months by using turnstile data from the official MTA site. We create a user-friendly interface that allows for the analysis of the busiest stations in NYC by time of day and how this changes as Covid-19 continues to affect everyone's lives. Our interface provides users a convenient way to understand which NYC subway stations are crowded during what time of day. As a result, users are able to explore how Covid-19 has directly affected MTA ridership.
My project, Interactive Spotify (IS), aims to provide an interactive audio-visualization of the user’s personal Spotify data by exposing certain aspects of Spotify’s song data that are hidden in its main application, namely the audio features for a given song. Audio features are abstract categorical descriptions of songs, such as energy or danceability, that are expressed as a numeric value between 0.0 and 1.0. IS allows a user to concretely view these otherwise hidden audio features and compare and correlate them with the various playlists in the user’s library. Although these audio features are not visible in the main application, they are available through Spotify’s API, and Spotify does refer to these features in official blog posts about listening trends over time. The project is an audio-visualization because it uses not just a visual mode of display, but an aural mode as well – users can listen to songs in real time as they interact with them in the visualization.
An exploration of the connections between artwork from the Metropolitan Museum of Art and the Rijksmuseum in a WikiRace format
Most of the star maps available now only visualize the spatial distribution of the stars. Although some of them provide a fancy rendering of the celestial bodies, users can not explore the non-spatial features of the stars. With these thoughts and inspirations, we explored a novel method to visualize the star dataset. We made a codex for stars that are well documented in databases (such as HYG and Simbad) based on their features such as their visual magnitude, color, and spatial locations. Therefore users can have an overview of how stars are distributed on these features and explore stars according to these characteristics. Using interactive techniques, non-spatial features are related to spatial features. Users can further inspect the star’s spatial distributions and explore detailed information such as the telescopic images of the stars.
xGraph-2 is an interpretable machine learning system for explaining machine learning classifications, specifically for automated anti- money laundering (AML) classifications of accounts on the bitcoin blockchain. GraphX-2 uses a methodology inspired by the “prototypes and criticisms” strategy for constructing visualization systems in order to explain the causes of local classifications.
This work is based on the ICML 2017 best paper "Understanding Black-box Predictions via Influence Functions". The researchers open sourced their work, but there aren't any known visualizations built on top of it. Typically, to compute the influence of a training example on a test example, we'd have to retrain the model with that one training example left out. This gets prohibitively expensive. This paper allows us compute these influences much faster than constantly retraining. The output basically gives a text list of which indices of training images were most helpful and most hurtful for making a prediction on each image. Text-based output and analysis for an image-based net is not very intuitive, so my project creates an interactive visualization to explore this output visually.
Previous work on visualizing prize data such as the Nobel Prize Laureate data usually only shows limited aspects using a static visualization. To fill this gap, we propose an interactive visualization tool for better exploration of the Nobel Prize Laureate through matrix and Sankey visualizations. A survey was conducted to gather feedback directly from users and demonstrate the effectiveness of our visualization.
The scrolly-telling story of how COVID-19 has rocked US domestic airports, flights, and airlines.
For avid music lovers, listening to music can be a deeply personal or revealing activity. Using a year of Spotify streaming history as data, we attempt to uncover patterns and habits about a user's listening preferences. Through a series of visualization methods such as heat maps and bubble charts, we summarize a user's past year using their top artists, songs, and genres, and show how listening activity varies across the hours in a day and the time of the year.
This final project is an experimental visualization tool that aims at optimizing the game exploration process. By looking into the way players choose their favorite games, the team found that the players tend to choose a new game mainly by the impression of the pictures and short videos, without even realizing their own preference clearly. To help the players clarify their unrealized preference and provide them with a more controllable and more efficient choosing approach, this research is divided into three parts. The first part is designed as a five choice questions that help us define the question-taker’s profile based on the tags attached to the selected games. The second part is a tag report that works essentially as an interactive output of the previous question set. The third part focuses on the extension of game attribute representation with the form of short videos. With this prototype, platform developers will be able to optimize their user-oriented recommendation and remold peoples’ habit of selecting their interested games.
How do people describe happiness, and how does happiness differ among people of different demographics? In this project, we design and implement a narrative visualization to help us and users of the visualization to answer these questions. Using HappyDB, a collection of 100,000+ responses to the question ”what made you happy today?” known as happy moments, the visualization takes the user through a personal narrative experience by asking the reader about their happiness and showing relevant results from the database. The visualization also presents general discoveries through a word cloud and a data exploration interface which make it convenient for the user to further explore the data at their own will.
We live in a world that is not just connected by technology and communication, but also by travel. The outbreak and rapid spread of the novel Coronavirus has caused many of us to miss the days when we could travel. It has reminded us that our world is more interconnected than ever before, and that the global tourism industry has impacts that go far beyond leisure travel; it carries significant upsides and downsides, with implications for local, national, and global economies, equity, sustainability, and security (especially biosecurity and cybersecurity). Given this, we were motivated to understand just how much the global tourism industry has grown over the past 25 years, and what have been some of the driving factors in that growth.
In this project, we develop an interactive webpage that visualizes the statistics of the MIT European Career Fair. It includes numerous visualizations about both ~3k student candidates and ~100 employers each year. We hope that this tool will allow both parties to inform themselves better about the fair to achieve their goals, which are to find jobs in Europe and to attract the best talents from top US schools, respectively.
The coronavirus (Covid-19) pandemic of 2019 came down on the world like an avalanche. Quickly, without warning or discrimination, schools, shops, restaurants and other businesses were suspended while most countries went under lockdown. People were forced to work from home, social distance themselves, and hospitals began to overflow with new patients. Markets plummeted and unemployment rates spiked. These are just a few (and the most immediate) of the economic and societal side-effects brought on by this pandemic. The short-term and long-term consequences of the lockdown measures are various and wide-reaching. Through data collected from Google Trends and Twitter, we look at the emotional implications of social distancing when it comes to the feeling of loneliness. Drawing from Catherine d’Ignazio’s theory on the role of emotion in data visualization we aim to relate our findings in a powerfully moving visualization.
Earthquakes, volcanoes, and tsunamis are the deadliest disasters in the world. News, books, and other media always broadcast terrible disaster scenes. With the rapid development of the Internet, can we combine data science technology with the latest data visualization tools to create an interactive, user-free disaster interface? To achieve this goal, we used Three.js, D3.js, WebGL, mapboxGL, react.js harp.gl, and leftlet to build the latest visualization platform so that people can understand the past disasters and serve people in disasters. Build a bridge of communication.