In machine learning, a hyperparameter is a parameter that is used to control the learning process. It is set by the user before the training process begins, and it does not change during training. Hyperparameters are typically used to control the complexity of the model, the amount of data that is used for training, and the learning rate.
The correct answer is: B. Parameters manually set by the user.
Option A is incorrect because parameters learned during training are called model parameters.
Option C is incorrect because model performance metrics are used to evaluate the performance of a model, but they are not hyperparameters.
Option D is incorrect because data transformation techniques are used to prepare the data for training, but they are not hyperparameters.