AI-woordenlijst
Het complete woordenboek van kunstmatige intelligentie
Polynomial Features
Feature engineering technique creating new variables by raising existing features to different powers to capture non-linear relationships in data.
Polynomial Expansion
Mathematical process transforming a feature vector into a higher-dimensional space by generating all polynomial terms up to a specified degree.
Polynomial Degree
Parameter determining the maximum power used when generating polynomial features, directly influencing model complexity.
Polynomial Overfitting
Phenomenon where a model with high-degree polynomial features overfits training data at the expense of its generalization capability.
Polynomial Regularization
Technique penalizing high coefficients of polynomial terms to prevent overfitting while preserving non-linear modeling capability.
Polynomial Basis
Set of polynomial functions forming a complete system to represent any function in an extended feature space.
Cross Features
Variables generated by multiplication or combination of existing features to reveal hidden patterns in variable interactions.
Chebyshev Polynomials
Family of orthogonal polynomials offering better numerical stability for approximating complex functions in feature engineering.
Polynomial Splines
Piecewise functions defined by polynomials joined continuously, offering increased flexibility to model local non-linear relationships.
Interaction Terms
Products of original features allowing the model to capture how the effect of one variable changes according to the values of other variables.
Linear Combination of Features
Mathematical operation creating new variables by weighted addition of existing features, often used as preprocessing before polynomial expansion.
Curse of Dimensionality
Problem where polynomial expansion creates exponentially more features, making learning computationally expensive and requiring more data.
Polynomial Feature Selection
Process of identifying and retaining the most relevant polynomial terms to avoid overcomplexity while maximizing the predictive power of the model.