Once we have determined whether a model is underfit or overfit over the given sample data, we must decide on how to improve the model's understanding of the relationship between the independent and dependent variables in our model. Let's briefly discuss a few of these techniques, as follows:
- Add or remove some features. As we will explore later, this technique can be used to improve both an underfit and an overfit model.
-
Vary the value of the regularization parameter
. Like adding or removing features, this method can be applied to both underfit and overfit models.
- Gather more training data. This method is a fairly obvious solution for improving an overfit model as it's needed to formulate a more generalized model to fit the training data.
-
Add features which are polynomial terms of other features in the model. This method can be used to improve an underfit model. For example, if we are modeling two independent feature variables,
and 
, we could add the terms 
as additional features to improve the model. The polynomial terms could be of even higher degrees, such as 
and 
, although this could result in overfitting the training data.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.