Book Image

The Kaggle Book

By : Konrad Banachewicz, Luca Massaron
5 (2)
Book Image

The Kaggle Book

5 (2)
By: Konrad Banachewicz, Luca Massaron

Overview of this book

Millions of data enthusiasts from around the world compete on Kaggle, the most famous data science competition platform of them all. Participating in Kaggle competitions is a surefire way to improve your data analysis skills, network with an amazing community of data scientists, and gain valuable experience to help grow your career. The first book of its kind, The Kaggle Book assembles in one place the techniques and skills you’ll need for success in competitions, data science projects, and beyond. Two Kaggle Grandmasters walk you through modeling strategies you won’t easily find elsewhere, and the knowledge they’ve accumulated along the way. As well as Kaggle-specific tips, you’ll learn more general techniques for approaching tasks based on image, tabular, textual data, and reinforcement learning. You’ll design better validation schemes and work more comfortably with different evaluation metrics. Whether you want to climb the ranks of Kaggle, build some more data science skills, or improve the accuracy of your existing models, this book is for you. Plus, join our Discord Community to learn along with more than 1,000 members and meet like-minded people!
Table of Contents (20 chapters)
Part I: Introduction to Competitions
Part II: Sharpening Your Skills for Competitions
Part III: Leveraging Competitions for Your Career
Other Books You May Enjoy

Trying different splitting strategies

As previously discussed, the validation loss is based on a data sample that is not part of the training set. It is an empirical measure that tells you how good your model is at predicting, and a more correct one than the score you get from your training, which will tell you mostly how much your model has memorized the training data patterns. Correctly choosing the data sample you use for validation constitutes your validation strategy.

To summarize the strategies for validating your model and measuring its performance correctly, you have a couple of choices:

  • The first choice is to work with a holdout system, incurring the risk of not properly choosing a representative sample of the data or overfitting to your validation holdout.
  • The second option is to use a probabilistic approach and rely on a series of samples to draw your conclusions on your models. Among the probabilistic approaches, you have cross-validation, leave-one...