03 : Generalization, Overfitting and Underfitting Flashcards
(5 cards)
1
Q
What is Generilazation?
A
Process of creating a model that is good at understanding new data because it learned overall patterns, not just memorized the specific training examples.
2
Q
Why Generalization important?
A
- Consistency : A well-generalized model performance consistently across both training and test datasets.
- Real-world predictions : Helps in making real world predictions rather than just fitting the training set.
- Overfitting & Underfitting : Poor generalization leads to Overfitting & Underfitting, making the model unreliable.
3
Q
Factors affecting generalization?
A
- Quality of data
- Size of dataset
- Model complexity
- Regularization techniques
4
Q
What is Overfitting?
A
Process of creating a model that perform great at training data but bad at new data.
5
Q
What is Underfitting?
A
Creating a model that perform on poorly on training data, that will also perform poorly on testing data too.
