Checking For Train, Test, Split Success

Author(s): Adam Ross Nelson

Originally published on Towards AI.

A look at ascertaining the success of a train, test, split

This article is a look at checking for a successful train, test, and split. Few tutorials discuss this step. The process of executing a train, test, and split procedure is necessary in machine learning and data science to avoid over-fitting and a range of other undesirable results.

A train test split avoids testing a model’s results and abilities with the same data used to train the model. In such a case, the model would likely achieve nearly 100% on any evaluation metric but then later perform poorly when it encounters previously unseen data.

Image Credit: Author’s illustration created in Canva.

If you’re at all like me, you occasionally miss out on sleep for over-thinking a recent project. Perhaps you lay awake because you’re just not sure there was more you should have done on a recent project. Otherwise, you might be stuck wondering if there was something different you could have done on a recent project.

First, this article will discuss what a successful train, test, and split is. Second, I also revisit p-values and the notions of randomness in statistics. Lastly, I will show code that readers may use to evaluate the success of train test splits in practice.

The documentation from SkLearn says… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI