What Are Data Science Challenges?

A data science challenge is a special kind of challenge for problems related to machine learning, data science, mathematics and statistics. They are typically longer in duration, usually at least a week, and often up to a month or more. The contestants are tasked with developing a data science solution that matches the given problem area and data as closely as possible. Some example cases might be the development of an image classification model, or to forecast data based on historical data points.

Data Science Challenge Types

MARATHON MATCH

A Marathon Match objectively scores with an automated scoring function that feeds the Topcoder live leaderboard. Types of submissions include actual code that implements a solution or a result set that represents the answer to a problem. Matches a configured to accept certain types of submissions. A set of “scorers” are configured for each match to run/validate the submissions and provide scores and feedback to the competitors. Scorers can be run by Topcoder or they could be external to the Topcoder Platform.

MINI-MARATHON MATCH

A Mini-Marathon Match is a version of the marathon match with a shorter duration, smaller prize purse, and is more straight-forward but with all the same structure and rules of a regular marathon match.

DATA SCIENCE SPRINT

A Data Science Sprint is a series of smaller more rapid Data Science challenges. It allows competitors to use any language and libraries to write their solution. Currently, these are run as Code challenges.

DATA SCIENCE IDEATION

A Data Science Ideation looks to discover or learn new approaches and discover data and ideas from the community.