Data Representation in Machine Learning

Definition

Applying linear and logistic regression to a real-life problem is very limited.

We want to a non-linear function that can describe our data, by applying a non-linear transformation to the data matrix \(\mathbf{X}\) prior to the fitting step.

By transforming our data matrix \(\mathbf{X}\), we are obtaining a data representation.

Example

Polynomials in polynomial linear regression.

References

Latent Factors and Factor Analysis Radial Basis Function Networks (RBF)