Kaggle Kernels were formally referred to as Scripts. The kernel simply refers to the Kaggle’s analysis, coding and collaboration product. According to the founder Anthony Goldbloom, this new name is more fitting because kernels are no longer short scripts that help in performing small tasks. They have been improvised and enhanced to create a product that is a combination of code, input, and output all stored together to cater for any version you choose to use. Since kernels enable you to store different attributes together, they are naturally reproducible, very simple to learn and extremely easy to share.
In Kaggle the kernel is an indispensable tool, foundation and core of your work as it contains the code required for analysis. Kaggle kernels contain code that helps make the entire model reproducible and enable you to invite collaborators when needed. It’s a one key solution for data science projects from code to comments and from environment variables to required input files. In future, we hope to see kernel’s integration with our local machine environment and more of an open collaboration tool where friends, employees, and teams can come across the world and contribute. We have also seen Kaggle kernel use in academic papers and research.
The indispensable Kaggle kernel runs exclusively on docker containers. For each Kaggle user, a kernel works by mounting the input into the containers that feature docker images that are already pre-loaded with the most common data science libraries and languages. In plain terms, a kernel is essentially a notebook or a script with data. It offers a number of advantages including, the containerization comes in handy in allowing contributors to set up their Kaggle projects, the users do not have to download data because it is already mounted in the docker container and the kernel code can be easily shared. It also offers transparency of shared code and makes it more accessible for beginners and experts alike.
How to Take Advantage of Kernels
Go through the top ranking kernels on a regular basis to get an idea of the thought process of other Kaggle contributors. Kaggle is a platform for learning; you should take advantage of any information and ideas you can get to improve yours skills. Overtime you will realize that you can easily increase your chances of winning if you use and combine the ideas. Use these kernels to improve your skills set and advance your knowledge in data sciences.
Kernels are a great way to boost transparency and also share code with other Kaggle contributors. This eliminates the chance that any contributor is left out of a piece of code buried somewhere else, it levels the playing field for all who like to learn, explore and improve their data science skills.
Qualities of a Good Kernel
On Thursday of every week, the Kaggle team comes together to select the best kernel using datasets available on the platform for the previous fourteen days. When choosing a winning kernel, there are two main considerations – Quality, the code of high quality consists of both a code and narrative that shares invaluable insights and also makes an impact that helps other Kagglers to learn, and the Quantity, the number of comments, UpVotes, and forks (the copies of your kernel made by other Kagglers). The winner is revealed on social media weekly using the hashtag #KernelsAward.
Publish Your First Kernel
Ask yourself what insights or perspectives are you trying to educate the data science community about. Be creative, do you have something unique to share, a tool, some perspective, or new ways to explore data. Feel free to create a tutorial that helps you share your knowledge and expertise, visualize data or reveal the hidden patterns. Here are examples of some great kernels that have been featured on Kaggle – Generation Unemployed? Interactive Plotly Visuals by Anisotropic using data from World Bank youth unemployment rates, Analyzing soccer player faces by SelfishGene using data from the Complete FIFA 2017 player dataset, and Traffic Fatalities in 2015 by Abigail Larion using data from 2015 Traffic Fatalities.
Now, the next step is to publish your own kernel. Simply click on New Kernel then select the data sources to use and a notebook or script to use. Publish both your narrative and code. Make sure to make your kernel public so other users can see and play with it. It will also get their feedback, comments, forks, and UpVotes, and you are automatically in the run to be selected as a winner.
The next step is to broadcast and publicize your work; it does not stop at sharing your kennel to the public. One of the most reliable ways to demonstrate the impact of your kernel is by sharing it widely within the Kaggle community. Broadcasting entails encouraging your connections on Kaggle to fork your kernel, UpVote, and comment and write a post and blog about it. Some effective ways to broadcast your kernel include sharing on social media accounts with proper hashtags like #Kaggle #KernelsAward etc.
You should also share your insights and motivations to write your kernel on a blog post and then share it with Kaggle and social media community.
Since it’s all about learning on Kaggle, you do not have to participate by creating your Kernel. You can also participate by being an active spectator. Keep up to date by checking out the latest kernels then comment and UpVote the ones you like. Fork your favorite kernel and see what changes you can make to improve its efficiency and performance. By doing this one day you will able to publish your own kernel.
I’ve recently published a book – Kaggle for Beginners – I hope you will enjoy it.