Skip to content

Latest commit

 

History

History
439 lines (341 loc) · 16.5 KB

instructions.md

File metadata and controls

439 lines (341 loc) · 16.5 KB
Raw

Project Overview

Over the rest of the semester, you will work towards creating an interactive, data-driven story. To extract value from large and complex datasets, data visuals have evolved over the past decade from static charts and graphs to interactive and immersive visuals that tell a story. This allows the audience to modify elements of the data being presented and manipulate the graphical representation. Static charts and graphs do not have the capability to adjust the visual, such as hovering, sorting and scaling. Interactive visualizations allow users to generate transformative insights, identify relationships, view trends and create meaningful stories through data.

Data storytelling is the process of describing data through visualizations by building a compelling narrative around datasets. This adds meaningful context to the data and helps the audience easily understand the information.

In your work as data scientists, in addition to doing modeling and machine learning work, you will be responsible (either individually or as part of a team) for providing the following as part of a project:

  • Findings: what does the data say?
  • Conclusions: what is your interpretation of the data?
  • Recommendations: what can be done to address the question/problem at hand?

Your narrative will focus primarily on the first two above. The narrative should allow your audience to be able to understand the topic you are analyzing, presenting and discussing.

We do not expect you to tackle and solve a large scale problem, that is not the objective. However, you are welcome to offer ideas and proposals on the topic you present. The objective is to find a topic of interest, work with the data, and present it to an audience that may not know very much about the subject using a data-driven narrative.

This is your chance to show creativity and have some fun!

Inspiration

The following are exemplary projects from prior years:

You can also take a look at the Juice Analytics 20 Best Data Storytelling Examples for more inspiration.

Milestones overview

There are four major milestones. Click on each milestone for the specific instructions.

  1. Dataset & topic proposal: you will propose dataset(s) to be used, the source, and the reason for using this dataset. You will also provide a short explanation of what you intend to do. The instructional team has the right to reject the dataset if it will not allow for a successful and comprehensive narrative.
  2. Work-in-progress milestone: work in progress submission where most of the visualizations and narrative are developed in some form but not finalized.
  3. Final milestone: the interactive narrative with all of its final components in a working website.
  4. Poster presentation: the technical document that supports your narrative.

Deliverable

The project will be delivered through an interactive self-contained website. The website and all of your code work will be done within the team GitHub repository unless otherwise specified (similar to how you have been doing assignments.) The same repository will be used throughout the life of the project.

Think of the entire project as several parts (for different audiences):

The Story

  • The visual narrative (a website): this is the end product and your audience is a general audience that wants to learn something about the topic you are presenting.

The Technical Approach

  • The code: this is the how. This includes all the code used to process your data, build the visualizations, and produce the website.
  • The poster: this is a technical summary. The audience is interested in understanding what you did and why you did it to get to the end result.

Overall Requirements

Important

Please read this section carefully. There are requirements for different aspects of the project.

Visualization

Your narrative must include the following elements:

  • At least five unique views, where a view is a visualization type. A view may also contain multiple plots. For clarification, the following are two examples of what can be considered 5 views:
    • One choropleth, one bubble-plot, one box-plot, one parallel coordinate plot, and one linked scatter-plot/histogram. This is five views and six plots.
    • Two choropleths, two bubble-plots, one visualization with a 3x3 multi-variable pairplot with histograms on the main diagonal (which is essentially a scatter-plot/histogram), one scatter plot, three linked boxplots/scatterplot, two parallel coordinate plots, and two histograms. This is five views and twenty one plots.
  • At least one of these views should be an attempt of an innovative view:

What is innovative

  • An extension of an existing visualization type
  • A novel visualization type
  • Something creative, novel, difficult, interesting, and generally beyond typical. It’s probably innovative if:
    • You are surprised that you were able to pull it off
    • You are pleasantly surprised by the insights that you’re providing and we learn something unexpected and novel about your data,
    • It’s something hard to get working and requires a deep understanding of graphic objects

What is not innovative

  • An interactive visualization
  • Something may not be innovative enough if:
    • Getting it to work is trivial
    • It’s a very common plotting method
    • It can be done with 5 or 10 lines of code
    • We can google it and quickly find a template of the code that generated the plot
    • It was done with plotly express
    • If it is fancy and difficult, but doesn’t tell us anything, or doesn’t make us think about your data, or doesn’t provide some insight about your data in alignment with your narrative, it might be innovative, but it will get penalized under the “graph for graph’s sake” rubric
  • At least two linked views. A click/hover/selection interaction within one view must trigger a change in a different view.
  • At least one view with Interactive tooltips that are shown when users hover over marks.
  • A formatted and/or interactive data table that presents part of your data in a pleasant and interactive way.

Important

All views and visualizations must have a purpose! Do not create a views just for the sake of meeting a requirement.

Design

You must apply best practices in design:

  • You narrative must have custom theming, look and feel.
  • Choose appropriate color schemes and keep them consistent.
  • Add appropriate narrative, captioning, etc.
  • Your website and visualizations must be as self-documenting as possible, with appropriate labels for panes, axes, and widgets, a legend documenting the meaning of visual encodings, and a meaningful title and description.
  • Your design choices may be creative and unconventional as long as they are serving a purpose.

Website

  • The website must:
    • Be self contained and accessible via an index.html file.
    • Have navigation tools, allowing the user/visitor to explore all the contents and always know where they are. The navigation implementation will depend on how you structure your site.
    • Show the title of your visual narrative, and the team members.
  • You are not required to serve or deploy your website for the project. You may do so at the end of the semester.

Code and Workflow

We expect to see best practices in team development and coding. All team members are expected to contribute to the development of the project.

  • All team members must commit
  • Use best practices in team workflow (branches, pull requests, merges)
  • Make sure that your code is well-organized and easy to read.
  • Use separate files for visualization and workflow components
  • Use functions to promote code reuse.
  • Don’t work with a messy repo and then try to clean it up before the final milestone. Comment your code early on!

Repository

  • Your repository must be well organized.
  • The website must be in it’s own subdirectory.
  • The instructors must be able to replicate all the pieces of your work.

Tools

Allowed

Your efforts will be purely on the data manipulation, analysis, and presentation using R and Python with their appropriate packages that wrap D3 or its variants and other front-end development tools (JavaScript, D3, HTML, CSS.)

For the website, you can choose the framework/tool to build your website (Rmarkdown, Quarto, or editing HTML directly.)

Warning

If you intend to use a package or tool not discussed in class, you must get prior approval from the instructional team.

  • You may choose to develop the whole narrative directly in D3 and JavaScript, however, in class we will be focusing on showing you R and Python wrappers to D3 and JavaScript that will make generating visualizations easier.

  • You may use CSS frameworks, such as Bootstrap, Materialize, or Distill and include external libraries (jQuery, leaflet.js, moment.js, etc.).

Not allowed

  • Any server side or custom backends (Node.js, Python, etc) and database systems, such as Postgres or MySQL.
  • Highcharts JS
  • Shiny

Evaluation

The project will be evaluated using the following high-level criteria:

  • Visual design: effectiveness of visualizations and interactions
  • Level of technical difficulty and quality of implementation
  • Whether the visual narrative answers a question, tells a story, and addresses the goals and requirements
  • Quality and clarity of your writing and overall presentation, including your own visual style

See the grade brackets in the syllabus for a detailed description.

Dataset and Topic Proposal

Proposal Submission

Important

Please submit only one form per team. Please coordinate amongst your teammates. You must be logged into GU Google and the respondent’s email address will be logged.

Please fill out this Google form with the following information:

  • Your team number as defined in Canvas
  • What dataset(s) are you planning to use? You must provide source URLs so we can take a look, and a brief description of the datasets. Make sure that the URLs for your data sources are correct and functional.
  • Why do you want to use this data? What do you wish to explore? (We don’t expect this to be a final answer, but at least show directional thinking.) ::: {.callout-tip} You may want to write this in a text file before submitting the form and then cut/paste your responses. :::

We will reject dataset/topic proposals that:

  • Do not have URLs to the source (or if the source is not functional)
  • Do not provide a thoughtful answer to the why and what
  • Intend to use trivial datasets that are not rich enough to allow you to do a comprehensive project

Dataset and Topic Requirements

  • The project topic is up to you and you can use any publicly available dataset(s).
  • You are allowed to use Kaggle datasets. However, keep in mind that Kaggle datasets are primarily designed for machine learning and modeling. While you can build visualizations from these datasets, you may find there may not be enough data points and/or variables or the data itself is masked or obfuscated. Any dataset with obfuscated fields or that has a large body of developed code/notebooks in Kaggle will not be allowed.
  • No proprietary datasets are allowed. All data must be available and accessible from public sources. Data behind a login is acceptable as long as anyone else can access the data (with their own login.)
  • You must attempt to use multiple datasets then you can join or layer them in a way that makes sense. Pick something that interests you.
  • Groups may use the same dataset(s) with the condition that each group creates unique analysis and visualizations.
  • Ideally, your dataset(s) should include:
    • Both qualitative and quantitative data
    • A time element
    • A geospatial element
    • A text element
    • The ability to be transformed into a graph/network
  • You may not use any of the following datasets:
    • New York City Taxicab
    • Airline Delays
    • Amazon Reviews
    • Iris dataset
    • Penguins dataset
    • Any datasets used in labs, lectures, or homework assignments
    • COVID related

There are many sources of available datasets. Please search and think beyond the obvious places. Here are some suggestions:

Work-in-progress Milesone

Expectations

Review the minimum requirements and what tools are allowed.

Important

This is not a first or rough draft. Even though this is a work-in-progress submission, we expect to see a good amount of effort to get here. Your project should be over 50% done at this milestone.

  • There should be a scaffolding of the visual narrative with some indication of where the story is going.
  • Most visualizations should be developed and some form. It is OK if all the elements of a visualization and story are not fully developed.
  • You can use placeholder text
  • If there is a placeholder for a visualization as part of the narrative, indicate what it is.
  • There should some form of custom theming, look and feel. Defaults will not be accepted.
  • Your deliverable must be clean and professional even if details are missing. No sloppy work will be tolerated.

Submission

Create a repository release with a tag called v0.1-wip and commit/push by the due date. Tagging a release allows you to continue the development, even before the mid-point is graded. See instructions for creating releases on Github This is in line with professional standards when using git.

Final Project Milestone

Expectations

Review the minimum requirements and what tools are allowed.

The visual narrative will be in it’s final form. All visualizations, narrative, navigation and theming are complete.

Submission

Create a repository release with a tag called v1.0-final and commit/push by the due date. See instructions for creating releases on Github.

Poster

The poster becomes the technical document and the requirements will be provided shortly.