Who are you¶
I am a data science student¶
You can learn about data science by signing up to ongoing and past data challenges on the RAMP Studio website. Sign up for the site then choose a theme or a data domain to find a challenge of interest to you. Find an open event for your challenge of interest and sign up. See Using RAMP starting-kits for more details.
If the challenge is in “open leaderboard” mode, you can browse the code of all the submissions, including the best ones submitted by top students or professional data scientists. You can also download and play with the RAMP starting kits, independently of the data challenges on RAMP Studio.
I am a practicing data scientist¶
If you have a dataset and a predictive problem, you can build your own workflow using the ramp-workflow library, following examples from RAMP starting kits. You can then train and test your models locally and log both the models and scores in a simple file structure. If you want to collaborate with your fellow team members, you can use git and upload your RAMP starting kit to GitHub.
I am a data science teacher¶
If you would like to use one of the existing RAMP challenges with our RAMP server system, for teaching purposes, ask for a new event at RAMP Studio. You can browse the existing challenges by the data science theme you would like to focus on in your course, or by the data domain you would like to apply data science to.
If you have your own data set and a predictive problem, you can prepare a RAMP challenge your course. See Build your own RAMP challenge for more information.
I am a machine learning researcher¶
You can benchmark your new algorithm against all previous solutions submitted for a specific data challenge on RAMP Studio. You can start by downloading the starting kit that you would like to use for benchmarking, from the GitHub repository ramp-kits, and test your algorithm locally. You can then sign up at RAMP Studio and register for one of the events corresponding to the challenge you choose. You can then submit your algorithm as many times as you want and you will have access to the public leaderboard score described in the starting kit.
If you register with us for an official benchmarking, we will provide you a private test score for a small number of submissions of your choice, at a date of your choice (but only once).
I am a researcher in a domain science¶
If you have a predictive problem, you can submit it as a data challenge to incite data scientists to solve your problem. First, build your own workflow using the ramp-workflow library, following examples from RAMP starting kits, then contact us so we can upload it to RAMP Studio. We can then organize hackathons or longer data challenges, and use the problem in a classroom setting. We may also automatically benchmark the thousands of models that are already in the platform. You can also contact us if you have problems setting up a challenge using RAMP workflow.