除了这三种模型之外，还有其他模型有hyperparameter吗？

问题如下：

What is the difference between a hyperparameter and a model parameter? Provide an example of a hyperparameter in:

a. Ridge regression

b. The construction of a decision tree

c. The use of SVM where perfect separation is not possible.

选项：

解释：

A parameter is estimated by the algorithm for each dataset (such as the intercept and slope terms in a regression model), whereas hyperparameters are usually choices made based on existing knowledge and then imposed on the model a priori by the researcher. Optimal values for the hyperparameters can be determined using a gridsearch over plausible ranges for the hyperparameters using a validation sub-sample.

a. The hyperparameter in a ridge regression is usually denoted by and controls the relative weight assigned to the model fit (residual sum of squares) relative to the shrinkage term (the squared sum of the slope parameters).

b. When constructing a decision tree, hyperparameters can be used to control the tree to prevent it from overfitting—for instance, by specifying a minimum information gain before allowing the tree a further split, or the specifying the maximum permissible number of branches.

c. When using support vector machines, if a perfect separation is not possible, a hyperparameter can be used to control the relative weight given to model fit (how close the points are to their centroids) and the extent of the misclassifications.