With a parametric algorithm, the model is specified by the user and the algorithm determines parameters to plug into that model.
Parametric algorithms are commonly used in machine learning and statistics for tasks where the underlying data distribution is known or can be assumed. They are typically more efficient and require less data to train compared to nonparametric algorithms. Parametric algorithms are appropriate whenever the supplied information about the model is certain or probable in advance. They generally yield better results than nonparametric algorithms provided the assumptions made are correct.
A classic example is linear regression: Here, the relationship between the predictor variables and the target variable is assumed to be linear. The algorithm then determines the coefficients that best fit the data. If the linear assumption is correct, the parametric model will perform well.
However, if the assumptions made are incorrect, the model’s performance can suffer. In such cases, nonparametric algorithms, which make fewer assumptions about the data, may be more appropriate.
Parametric algorithms are an essential tool in machine learning and statistics for building models that can leverage prior knowledge about the data distribution to make accurate predictions.
- Related terms
- Nonparametric Algorithm