✍️ Notes on Getting Started on Kaggle
These notes have been taken from YouTube videos on Kaggle’s channel
- Kaggle Lingo
- Kernels = Jupyter notebooks and raw Python/Markdown scripts
- Kerneler = a bot that will write a kernels for any dataset that you upload
- Cells = in notebooks, cells are editable blocks of text or code
- Docker = types of containers (file systems) lets you run code written on Kaggle on your local computer
- Commit = commiting code runs it from top to bottom and saves static version you can refer back to (like a save file in a video game)
- Competition Types
- Featured : the most competitive with the biggest prizes. Often hosted by corporations looking to improve their current approach.
- Analytics : Not supervised machine learning. Designed to help nonprofits with tricky problems they’re facing.
- Research : More theoretical problems based on active areas of research (academic & industrial).
- Getting Started : Designed to be your first or second competition.
- Playground : Especially interesting problems that are fun to solve or are seasonal competitions.
- InClass = Lets you host your own Kaggle competition. (They don’t have to be for an actual class, thats just why it was started)
- Leaderboard = The place where all the scores for entries to a specific competition are shown.
- Public : THis is the one that you see during the competition
- Private : You only see this after the competition ends. Usually scored on a different part of the data.
- Shakeup = Changes in ranking when the private leaderboard is shown at the end of a competition. Usually people move down the ranks due to overfitting.
- In the money = people at that rank will win money if they still have that rank on the private leaderboard after the shakeup
- Leakage = When information about the target variable is accidentally included in the training data
- ex : you’re trying to predict if someone will get an illness but you don’t remove information about whether someone was treated for that illness
- Kaggle Progression System
- Novice : There is a checklist of things to do to become a Contributor
- Contributor : your votes count for awarding other people medals
- Expert : you get ranked separately in three catagories: discussion, kernels and competitions
- Master
- Grandmaster
- Medals = are awarded for having your content upvoted or being near/at the top of the leaderboard for competitions
📚 Resources Used in Researching Kaggle