Post project ideas, recruit team members, team up with talented people! Where UC Berkeley students come to put data to good use. Please make sure to make an account with your Berkeley email. To post a project: Navigate to the relevant category and click "Create a new page" To join a project: Scroll to the bottom of the project page you're interested in and click "Add comment" For more detailed instructions go to discoveryexchange.org
- How to Use the Discovery Project Exchange
- Project Template
- Formation Template
- Questions & Answers
- Project Formation
- Tony Hawk AI Skater [OPEN]
- [Project Boom Web] User Engagement Study and Establishing Effective User Flow
How to Use the Discovery Project Exchange
Click here to see actively recruiting projects! Read on for instructions:
Functions of Discovery Exchange
To facilitate the formation of groups around the shared goal of completing a specific data science-related project.
To gather questions and answers on all things related to your summer passion project, project management, data science best practices, etc.
To announce opportunities to participate in workshops, mentorship and consulting opportunities.
Team formation happens in the Project Exchange
You can build a data science team in 2 ways:
- Build a project idea from scratch
- Post a project idea already designed
- Join a project
Build a Project Idea
Interested in different topics, but don't have a project idea quite yet? Navigate to the "Project Formation" section and copy the "Formation Template". Create a new page under "Project Formation" and post about project ideas that you may have!
Post a Project
Here are the fields to include:
- Problem Definition - try to be specific and concise
- Data - do you have it already? where are you going to find it?
- Deliverable - what do you hope to achieve/build? what insights do you hope to gain?
- Timeline - what are some key checkpoints that you need to get through to complete your project? by what date?
- Teammates (optional) - who is currently on the team? who are you looking to join your project!
Once you have posted a project, be sure to check your page often to see who has expressed interest. Confirm with the respondents that you want to work with and schedule your first meeting. Good luck!
Join a Project
Browse the list of posted projects and reply to a couple of the posts that most interest you. Once the original project poster confirms with you, schedule your first meeting. Check back often as people will be posting projects all through the semester!
Check out presentations from some amazing data science projects that were founded through Discovery Exchange!
[OPEN/CLOSED] Project Title
About me: My name is [name] I am a [graduate/undergraduate] student looking to build a [ML model and web-based visualization] that I can [put in my portfolio].
Problem Definition - Given the huge problem [something] I want to see if we can [predict/get some insight] on [something else]
Data - We'll be using the [dataset] dataset which can be found here [link]
Deliverable - I hope to build a [Dash app] and write a blog post of our process and results
Timeline - [We'll need to have an initial analysis of the data done by June 15, a first pass at the model by June 1, and a working skeleton of the app by July 15. Then we can go back and refine those elements and start working on the written report.]
Team - I am currently working [alone, but am excited to meet potential team members!].
Looking for some [fun and enthusiastic] people to work with. Comment if you're interested!
Project Idea: [Your Idea Here!]
About Me: My name is [Your Name!] and I'm a [junior] studying [Data Science]
Areas of Interest: [COVID-19, Social Good, Humanities]
Platforms/Languages of Interest: [Machine Learning, Computer Vision, Data Visualization, etc.]
Message to the Group: [I'm super excited to get to know everyone! I have an idea for a project relating to COVID-19 but not sure where to start, anyone interested?]
Questions & Answers
Comment on this page with any questions you have and we'll update the FAQ
Use this section to form different projects!
Post your Covid-19 related data science project idea here!
Post your economics related projects here!
Post your engineering related projects here!
Tony Hawk AI Skater [OPEN]
About me: I'm Alex, a graduate student who's been working on a personal project synthesizing ML/reinforcement learning and the Unity Game engine to teach a skateboard how to ollie, kickflip, etc.
Problem Definition - Teach a skateboard how to do skate tricks with reinforcement learning
Data - No dataset needed!
Deliverable - Hope to post a video showcasing the behavior of the skateboard and maybe write a blog post explaining the process.
Timeline - I already have the Unity assets created and some basic behavior encoded to control the skateboard manually; over the next few months, we can learn how to use the pre-built Unity MLAgents package to control the skateboard with reinforcement learning and potentially include alternative methods too (evolutionary algorithms, etc).
Team - Currently working alone, but anybody can jump onto this project! I can take the time to introduce basic Unity and Deep Learning concepts to you too; a lot of the details are abstracted away in Unity, so beginners totally welcome.
Comment if you're interested!
[Project Boom Web] User Engagement Study and Establishing Effective User Flow
[OPEN] User Engagement Study and Establishing Effective User Flow
About me: My name is Minos Park. I am a recent undergraduate student working with global team of students building a supersonic aircraft attempting to break the world record. Check our website on https://theprojectboom.org/
Expected Project Outcome:
- Website (https://theprojectboom.org/) metric improvement, to drive more user engagement
- Implement an effective user flow based on the objective guidance from Project Boom lead
From the commencement of the project, duration of each phase is listed here:
1 week - Review existing data and research further data needs; which metrics to improve and how?
1 week - Choose and deploy an appropriate data platform/pipeline
2-3 weeks - Setup A/B tests/do modifications on the website based on the early research
1 week - Write interim report, iterate tests
variable weeks - iterate tests
1 week - Write a final report, which can be added to your portfolio
Team: You'll join the project as a member, under the marketing team.
- How to define and choose appropriate metric to measure the changes in user behavior
- Topics on user behavior analytics
- Conducting A/B Tests
- Experience with various analytics platforms - e.g. Google Analytics, Mixpanel, Segment, etc
Resources (more available):
- Google Analytics
Post your health-related projects here!
Post your humanities/social sciences related projects here!
Separating the Senate [OPEN]
About me: My name is Alex, a graduate student who is looking to use PCA/other statistical tools to quantify partisanship in the Senate, inspired by this (https://nicogj.github.io/post/2018/02/13/senate-polarization.html).
Problem Definition - Given the increase in political polarization and partisanship in Congress and outside of it, how can we quantify these differences? How can we quantify the change in polarization over time?
Data - We'll be using senator vote records scraped from the internet that I have already collected and cleaned (though it can totally be extended!)
Deliverable - I hope to create a set of visualizations and write a blog post of our process and discussion about our results (since I am not a Poli Sci student and I imagine most of you aren't either, this can totally be in the realm of speculation.
Timeline - N/A yet
Team - I am currently working alone, but am excited to meet potential team members! Comment if you're interested :)
Post your policy projects here!
Post projects here if they do not fit in other categories!
Data and happiness [OPEN]
About me: My name is Alex, a graduate student who would like to be happy :))
Problem Definition - can the act of "measuring" and observing data about our mental state affect our perceptions and help us be "happier"? (this is very exploratory)
There's this saying that "you can't manage what you can't measure." Most people don't actively manage or measure their mental/emotional states because, how would you even do something like that? Our inner lives are incredibly fluid and subject to so many external factors, and for extreme cases, is subject to biological factors that we cannot control of in the first place.
However, there may be some things that we can manage. Cognitive behavioral therapy, mindfulness practice, religious practice, etc. can be applied during mild fluctuations of mood and emotion to help us get through rocky times. Even something as simple as scrolling through my camera roll and remembering good times can pull me out of a funk and prevent ruminating too long on the negatives. Is there a data-driven way we can measure and reflect on our mental states?
Data - We'll be collecting data ourselves with a very simple mechanical counter. Click it when you feel a moment of happiness, and record the total number of clicks each day together with a (very) short journal entry (we'll be doing some NLP on the written portions and pair it with the self-reported "happiness" clicks).
Deliverable - Not sure yet! I'm thinking some use cases could be checking our happiness graph to reflect on moments of high happiness or particular struggles, comparing a machine-reported assessment of our mood (sentiment analysis) with our self reported mood, etc.
Timeline - Not sure yet! Could easily be a very long term project.
Team - Anyone who's interested in mindfulness, mental health, psychology, being happy, or the way data affects the way we perceive the world and ourselves.