Types of Kaggle Competitions
I am going to try to use Kaggle competitions to get better at machine learning. There are Kaggle competitions of varying difficvulty, described below.
Kaggle competitions are challenges in which data scientists and machine learning engineers compete to create the best models for solving specific problems or analyzing certain data sets.
Coursera
Kaggle distinguishes between 4 different types of competitions in terms of their difficulty. I am creating this note to remind myself of the levels of difficulty and to provide links to the competitions / datasets when I want to return focusing on learning Machine Learning. I don't hope to win any of these competitions, or get close to winning, but I instead just want to try to learn about machine learning to maybe implement some techniques in later projects.
Types of Competions
Getting Started
Getting started competitions are the easiest competitions on Kaggle that are designed for people new to mcahine learning. They have a rolling leaderboard. They offer no prizes or points. Here is a link to the getting started competitions.
Playground
Playground competitions are one step above Getting Started competitions in terms of difficulty. Here is a link to the Playground competitions.
Research
Research competitions feature problems that are more experimental than featured competition problems. They usually do not offer prizes or points, but they offer an opportunity to work on problems which may not have a clean or easy solution and which are integral to a specific domain. Here is a link to the Research competitions.
Featured
Featured competitions are designed to provide challenges for competitors at all different levels of machine learning knowledge. Featured competitions offer a way to learn from the best in the field. Here is a link to the featured competitions.
Competition Formats
- Simple Competitions
- Simple competitions are those in which you can access the complete datset at the beginning of the competition, build a model in a Jupyter Notebook, generate a prediction file, and upload your prediction as a submission on Kaggle. Most Competitions Follow this format.
- Two Stage Competitions
- The challenge is plit into two parts, with the second stage involving a new test dataset that is released at the start of the stage. Elligibility for stage 2 typically requires making a submission in Stage 1.
- Code Competitions
- All submission are made from inside of a Kaggle Notebook, and it is not possible to upload submission to Competition directly.
- Thhese competitions are more ballanced, since all users have the same hardware limitations. These models tend to be far simpler than the winning models in other competitions.
Other Considerations
- You should read the rules tab of the competition before competing in the comptetion. Not doing so could result in your submission being invalidated at the end of the competition o you being banned from the platform.
- The description tab gives an introduction into the competition's objective
- The data tab is where you can download and learn more about the data used in the competition.
- The evaluation section describes how to format your submission file and how your submissions will be evaluated.
- The timeline tab has detailed information on the competition timeline.
- The prizes section provides a breakdown of what prizes will be awarded to the winner, if prizes are relevant.
Comments
There are currently no comments to show for this article.