🪴 Aradinka Digital Garden

Search

Search IconIcon to open search

overfitting

Last updated Nov 22, 2022

An overfitting model is the one that fits well with the training data, i.e. little or no error, however it does not generalized well to the unseen data

a model matches the training data almost perfectly, but does poorly in validation and other new data

overfitting capturing spurious (false) patterns that won’t recur in the future, leading to less accurate predictions

How to avoid overfitting?