Reputation: 23
For a KNN model, is the value of K inversely proportional to its capacity/complexity? How can one determine whether a curve is a training curve or a validation curve based on its Loss Vs Capacity graph? Thanks! ML newbie here.
Upvotes: 0
Views: 170
Reputation: 7399
K in KNN is a hyperparameter that determines how many (K) votes from the 'nearest neighbors' are supposed to be considered to make a prediction against a sample. What do you mean by the capacity of the model? Also, the larger K is, the more complex the model.
Upvotes: 1