We’ve built a network of real-time data and algorithms to make low-cost, high-quality predictions accessible and useful. Getting predictions is as simple as publishing data to our platform where competing, state-of-the art algorithms converge on optimized predictions.
Micropredictions LLC is a subsidiary of Intech Investments, a top five U.S. Equity fund manager by a variety of metrics. Micropredictions LLC is established to foster development of the world’s first microprediction network and its use by organizations large and small. We support the open source software development both directly, and through contributor incentives. We host the first node at www.microprediction.org, where you will find an API can be used to predict literally anything. We help companies or organizations get up to speed with usage patterns, including privacy preserving use of public prediction, and the techniques linking repeated prediction and business optimization.
In this project, student would learn how to write and deploy programs that directly address real-time operational problems. These would predict quantities such as electricity (https://www.microprediction.com/competitions/electricity) or cryptocurrency movements.
Students would also be able to create intelligent applications, such as a web page, that uses the turnkey prediction API to drive intelligence, thus pioneering new ways to exploit cheap prediction.
Course Info & Engagement Details
|Engagement Format||Live Case - Class Collaboration or Case Competition - This learning format allows educators to deliver experiential learning to students at scale. Students are often split into groups to work on a live case (or a series of cases) from a real host company that directly relates to key learning objectives.|
|Course||Spring Virtual Data Science Consulting Project|
|Meeting Day & Time||Monday, Wednesday, Friday: 11:00 AM MT - 11:50 AM MT|
|Student Time Commitment||8-15 Hours Per Week|
|Company Time Commitment||2 Hours|
Product Design & Development
Research & Development
Software Design & Development
There are currently no students assigned.
November 30, 2020
Deadline for Industry Partners to Apply to Participate
December 14, 2020
Project Scope Finalized
January 18, 2021
OFFICIAL PROJECT LAUNCH: We’ll find a time on this day to web conference you into our class to kickoff the project.
April 23, 2021
OFFICIAL PROJECT END: We’ll find a time on this day to web conference you into our class to close the project.
Key Milestones & Project Process
February 27, 2021 - Deep-dive into Micropredictions site and its features and functionalities
- What are the central objectives for the site?
- What features and functionalities are available?
- How does the site operate, and how does the site facilitate the user’s engagement with the open microprediction network? For instance:
- How does the user pull historical data and launch an algorithm?
- What data are available live stream? How is the data accessed, monitored and utilized?
- How is Python used in this context?
- As they proceed through the Python tutorials
- Create an identity (MUID)
- Enter a “z1” contest in a one-off fashion by estimating a univariate distribution
- Run the default crawler
- Create a data stream for others to predict, using an interesting source of live data that updates every 15 minutes (e.g. transport)
- Use an open source Python package to improve the prediction model in the crawler
- Modify the crawler to focus on certain types of streams
April 23, 2021 - Create and deploy a well-documented and useful quantile estimation open source library
- Research and collect background material and data sources for online quantile estimation found in open source world
- Write and deploy quantile algorithm for estimation
- Find relevant exogenous data for the deployed model, such as humidity impacting electricity consumption.
- Create user-friendly open source library framework and populate with background material and data sources
- Learn how to deploy a package to PyPI https://www.linkedin.com/pulse/you-love-your-algorithm-set-free-peter-cotton-phd/
- Open source library for online quantile estimation
- Blog article featured on Micropredictions website on the subject of online quantile estimation for online analysis and the new open source library
There are no resources currently available
Company Supervising Team
There are currently no supervisors assigned.
There are currently no supervisors assigned.