Sunday, May 15, 2022
HomeArtificial IntelligenceUtilizing Kaggle in Machine Studying Tasks

Utilizing Kaggle in Machine Studying Tasks

You’ve in all probability heard of Kaggle knowledge science competitions, however do you know that Kaggle has many different options that may show you how to along with your subsequent machine studying challenge? For folks in search of datasets for his or her subsequent machine studying challenge, Kaggle lets you entry public datasets by others and share your personal datasets. For these trying to construct and prepare their very own machine studying fashions, Kaggle additionally presents an in-browser pocket book setting and a few free GPU hours. It’s also possible to have a look at different folks’s public notebooks as effectively!

Apart from the web site, Kaggle additionally has a command-line interface (CLI) which you should use throughout the command line to entry and obtain datasets.

Let’s dive proper in and discover what Kaggle has to supply!

After finishing this tutorial, you’ll be taught:

  • What’s Kaggle?
  • How you should use Kaggle as a part of your machine studying pipeline
  • Utilizing Kaggle API’s Command Line Interface (CLI)

Let’s get began!

Utilizing Kaggle in Machine Studying Tasks
Picture by Stefan Widua. Some rights reserved.


This tutorial is break up into 5 components; they’re:

  • What’s Kaggle?
  • Organising Kaggle Notebooks
  • Utilizing Kaggle Notebooks with GPUs/TPUs
  • Utilizing Kaggle Datasets with Kaggle Notebooks
  • Utilizing Kaggle Datasets with Kaggle CLI software

What Is Kaggle?

Kaggle might be most well-known for the info science competitions that it hosts, with a few of them providing 5-figure prize swimming pools and seeing tons of of groups collaborating. Moreover these competitions, Kaggle additionally permits customers to publish and seek for datasets, which they’ll use for his or her machine studying initiatives. To make use of these datasets, you should use Kaggle notebooks inside your browser or Kaggle’s public API to obtain their datasets which you’ll then use in your machine studying initiatives.

Kaggle Competitions

Along with that, Kaggle additionally presents some programs and a discussions web page so that you can be taught extra about machine studying and speak with different machine studying practitioners!

For the remainder of this text, we’ll deal with how we are able to use Kaggle’s datasets and notebooks to assist us when engaged on our personal machine studying initiatives or discovering new initiatives to work on.

Organising Kaggle Notebooks

To get began with Kaggle Notebooks, you’ll must create a Kaggle account both utilizing an current Google account or creating one utilizing your electronic mail.

Then, go to the “Code” web page.

Left Sidebar of Kaggle Residence Web page, Code Tab

You’ll then be capable of see your personal notebooks in addition to public notebooks by others. To create your personal pocket book, click on on New Pocket book.

Kaggle Code Web page

This may create your new pocket book, which seems like a Jupyter pocket book, with many related instructions and shortcuts.

Kaggle Pocket book

It’s also possible to toggle between a pocket book editor and script editor by going to File -> Editor Kind.

Altering Editor Kind in Kaggle Pocket book

Altering the editor kind to script reveals this as an alternative:

Kaggle Pocket book Script Editor Kind

Utilizing Kaggle with GPUs/TPUs

Who doesn’t love free GPU time for machine studying initiatives? GPUs might help to massively velocity up the coaching and inference of machine studying fashions, particularly with deep studying fashions.

Kaggle comes with some free allocation of GPUs and TPUs, which you should use in your initiatives. On the time of this writing, the supply is 30 hours every week for GPUs and 20 hours every week for TPUs after verifying your account with a telephone quantity.

To connect an accelerator to your pocket book, go to Settings ▷ Surroundings ▷ Preferences.

Altering Kaggle Pocket book Surroundings preferences

You’ll be requested to confirm your account with a telephone quantity.

Confirm telephone quantity

After which introduced with this web page which lists the quantity of availability you’ve left and mentions that turning on GPUs will cut back the variety of CPUs accessible, so it’s in all probability solely a good suggestion when doing coaching/inference with neural networks.

Including GPU Accelerator to Kaggle Pocket book

Utilizing Kaggle Datasets with Kaggle Notebooks

Machine studying initiatives are data-hungry monsters, and discovering datasets for our present initiatives or in search of datasets to start out new initiatives is all the time a chore. Fortunately, Kaggle has a wealthy assortment of datasets contributed by customers and from competitions. These datasets could be a treasure trove for folks in search of knowledge for his or her present machine studying challenge or folks in search of new concepts for initiatives.

Let’s discover how we are able to add these datasets to our Kaggle pocket book.

First, click on on Add knowledge on the suitable sidebar.

Including Datasets to Kaggle Pocket book Surroundings

A window ought to seem that reveals you a number of the publicly accessible datasets and offers you the choice to add your personal dataset to be used along with your Kaggle pocket book.

Looking By Kaggle datasets

I’ll be utilizing the traditional titanic dataset as my instance for this tutorial, which yow will discover by keying your search phrases into the search bar on the highest proper of the window.

Kaggle Datasets Filtered with “Titanic” Key phrase

After that, the dataset is obtainable for use by the pocket book. To entry the recordsdata, check out the trail for the file and prepend ../enter/{path}. For instance, the file path for the titanic dataset is:

Within the pocket book, we are able to learn the info utilizing:

This will get us the info from the file:

Utilizing Titanic Dataset in Kaggle Pocket book

Utilizing Kaggle Datasets with Kaggle CLI Instrument

Kaggle additionally has a public API with a CLI software which we are able to use to obtain datasets, work together with competitions, and way more. We’ll be find out how to arrange and obtain Kaggle datasets utilizing the CLI software.

To get began, set up the CLI software utilizing:

For Mac/Linux customers, you may want:

Then, you’ll must create an API token for authentication. Go to Kaggle’s webpage, click on in your profile icon within the high proper nook and go to Account.

Going to Kaggle Account Settings

From there, scroll right down to Create New API Token:

Producing New API Token for Kaggle Public API

This may obtain a kaggle.json file that you simply’ll use to authenticate your self with the Kaggle CLI software. You’ll have to place it within the appropriate location for it to work. For Linux/Mac/Unix-based working techniques, this must be positioned at ~/.kaggle/kaggle.json, and for Home windows customers, it must be positioned at C:Customers<Home windows-username>.kagglekaggle.json. Inserting it within the unsuitable location and calling kaggle within the command line will give an error:

Now, let’s get began on downloading these datasets!

To seek for datasets utilizing a search time period, e.g., titanic, we are able to use:

Looking for titanic, we get:

To obtain the primary dataset in that record, we are able to use:

Utilizing a Jupyter pocket book to learn the file, just like the Kaggle pocket book instance, offers us:

Utilizing Titanic Dataset in Jupyter Pocket book

After all, some datasets are so giant in dimension that you could be not need to maintain them by yourself disk. Nonetheless, this is likely one of the free sources offered by Kaggle in your machine studying initiatives!

Additional Studying

This part gives extra sources in case you’re fascinated about going deeper into the subject.


On this tutorial, you realized what Kaggle is , how we are able to use Kaggle to get datasets, and even for some free GPU/TPU cases inside Kaggle Notebooks. You’ve additionally seen how we are able to use Kaggle API’s CLI software to obtain datasets for us to make use of in our native environments.

Particularly, you learnt:

  • What’s Kaggle
  • The right way to use Kaggle notebooks together with their GPU/TPU accelerator
  • The right way to use Kaggle datasets in Kaggle notebooks or obtain them utilizing Kaggle’s CLI software



Please enter your comment!
Please enter your name here

Most Popular

Recent Comments