Generalization (Over fitting and under fitting)

• Generalization refers how well a model trained by a training set predicts the right output for new instances is called generalization.

• Two problems associated with generalization:             

                                          - Over fitting

                                          - Under fitting

• Over fitting and Under fitting are the two biggest causes for poor performance of machine learning.

• Underfitting: it is the production of a machine learning model which is not complex enough to accurately capture the relationship between a data-set feature and a target variable.

• Overfitting: It is  when a model learns the detailed feature and noise to the extend that it negatively impacts the performance.