Overview#
Underfitting or High
Bias error occurs when a
Machine Learning model does not adequately capture the underlying structure of the
Training dataset.
Underfitting is when the Machine Learning model does not perform well on the Training dataset.
An Underfitting Machine Learning model is where some parameters or terms that would appear in a correctly specified model are missing.
Underfitting would occur, for example, when fitting a linear model to non-linear data. Such a model will tend to have poor predictive performance.
Solving for Underfitting#
A bigger
Machine Learning model (ie more
Hidden layers) will almost always just reduces your
Bias error without necessarily increasing
Overfitting (
Variance error), so long as you regularize appropriately.
Generally, using a larger Training dataset will NOT solve Underfitting
There might be more information for this subject on one of the following: