Overfitting vs Underfitting

Overfitting — at a glance

Category: Fundamentals · Difficulty: Beginner

Overfitting is when an AI model memorizes the training data instead of learning general patterns, making it perform poorly on new data.

Overfitting occurs when a model's hypothesis complexity exceeds the intrinsic complexity of the data-generating process, resulting in low training error but high generalization error. Formally, the model has low bias but high variance in the bias-variance decomposition.

Read the full Overfitting definition →

Underfitting — at a glance

Category: Evaluation · Difficulty: Intermediate

A phenomenon where a model is too simple to capture the underlying patterns in the training data, resulting in poor performance on both training and new data.

Underfitting occurs when a machine learning model is too simplistic or has not been trained sufficiently to capture the meaningful relationships in the training data. This results in high errors on both the training data and new, unseen data. It indicates that the model is not complex enough to learn the target function.

Read the full Underfitting definition →

Key differences

• Purpose: Overfitting is typically used for fundamentals problems, while Underfitting fits evaluation use cases.
• Complexity: Overfitting is rated Beginner; Underfitting is rated Intermediate.
• Definitions: Overfitting is when an AI model memorizes the training data instead of learning general patterns, making it perform poorly on new data. vs A phenomenon where a model is too simple to capture the underlying patterns in the training data, resulting in poor performance on both training and new data.

Frequently asked questions

What is the difference between Overfitting and Underfitting?

Overfitting: Overfitting is when an AI model memorizes the training data instead of learning general patterns, making it perform poorly on new data. Underfitting: A phenomenon where a model is too simple to capture the underlying patterns in the training data, resulting in poor performance on both training and new data.

When should I use Overfitting instead of Underfitting?

Use Overfitting when your problem matches its strengths: Overfitting is when an AI model memorizes the training data instead of learning general patterns, making it perform poorly on new data. Use Underfitting when A phenomenon where a model is too simple to capture the underlying patterns in the training data, resulting in poor performance on both training and new data.

Can Overfitting and Underfitting be used together?

Yes — many modern AI systems combine Overfitting and Underfitting to get the strengths of both approaches.

Is Overfitting better than Underfitting?

Neither is universally better. The right choice depends on data, latency, cost, and task. This page breaks down the trade-offs.