Parametric model in Machine Learning
Parameter is an internal configuration variable of a model. The value of the parameter can be determined from training data.
For example: Coefficient in linear and logistic regression, weight in neural network and support vector in SVM.
Parametric models are consist of number of fixed parameter. All the fixed parameters are independent of number of training instances.The number of parameter will be never changed if the volume of training data is increased.
Learning function is a mapping between input and output variable in machine learning.
In case of parametric model f(x) is always assume in a known form. For example, in Naive Bayes, it is assumed all the features are independent, in logistic regression it is assumed all the positive and negative variables are linearly separable.
Example of parametric model :
- Linear regression
- Logistic regression
- Linear SVM
- Naive Bayes
- Perceptron
Advantage: Simpler , Easy to train, take less training time.
Disadvantage: Assume the mapping function prior, suitable for less complex problem .
Reference: https://machinelearningmastery.com/parametric-and-nonparametric-machine-learning-algorithms/
