Non-parametric Does Not Mean "No Parameters"

What is the difference between parametric and non-parametric models? Parametric models make assumptions about the mapping function between inputs and outcomes. In these models, there is a finite and set number of parameters, which is used to make predictions. The number of parameters is set because changing the amount of data will not change the number of parameters required to make predictions. Examples of parametric models include Linear regression, Logistic regression, Linear Support Vector Machine, and others.

In contrast, non-parametric models do not make any assumptions about the mapping function between inputs and outputs. As such, non-parametric models are free to choose the function based on the data. Consequently, there are potentially multiple possible parameters. Thus, non-parametric does not mean "no parameters"; on the contrary, it reflects the possibility of multiple parameters. Changing the amount of data will likely change the number of parameters. Examples of non-parametric models include K-nearest neighbors, decision trees, inter alia.

Further particulars at reference 1, reference 2, and reference 3