Posted in Statistics, Total Reads: 624

Definition: Overfitting

Statistical models are often used to demonstrate the existing relationship between a number of parameters and the respective observations. Often when there are too many variables relative to the number of observations then often such a model shows the existing noise rather than the underlying relationship. Owing to Overfitting minor fluctuations in the data are also over exaggerated and shown.

A very simple example is if two variable can be easily explained with a linear relationship then the usage of more parameters can lead to the Overfitting of the model. Overfitting leads to wastage of resources and is considered undesirable.

Overfitting can be avoided by independent validation, splitting the sample and resampling the method used the data collection method should be well planned as per the literature to avoid Overfitting.


Looking for Similar Definitions & Concepts, Search Business Concepts