Underfitting
In plain terms
When a model is too simple to capture data patterns.
Going deeper
Underfitting happens when a model can't represent the underlying structure, performing poorly on both training and new data. It hasn't learned enough.
Under the hood
Caused by too little capacity, over-regularization, or insufficient training; the opposite of overfitting.
Examples
Too-simple models, high bias, poor train accuracy, insufficient features.
Think of it like...
Like trying to trace a winding road with a single straight ruler.