Too simple to learn.
Underfitting happens when a model is too simple to learn the patterns in the data, leading to poor performance on both training and new data.
Imagine trying to draw a detailed picture of a cat using only a thick crayon. You might get the basic shape, but you'll miss all the whiskers and fur details, so the drawing looks like a blob. That's underfitting — your tool (the model) is too simple to capture the important patterns.
Underfitting matters because it means your AI model is not smart enough to solve the problem, wasting time and resources. You often encounter it when starting with a very simple model like linear regression for complex data, or when training is stopped too early.
Many people think underfitting means the model didn't train long enough, but it's actually about the model being too simple, not just undertrained. Training longer won't fix underfitting — you need a more complex model or better features.
Underfitting occurs when a machine learning model is too simple to capture the underlying structure of the data, resulting in high bias and poor generalization. It fails to perform well on the training data and consequently on unseen data. This is often due to insufficient model capacity, overly strong regularization, or inadequate feature engineering.