Labs Flashcards
1
Q
Explain briefly K nearest neighbours
A
It is based on the idea that models that are close together in feature space belong to the same class
the fitting is mostly based on the idea of indexing all the sample
while the prediction computes the euclidean distance to the other points and then finds simply the K nearest neighbourhood
2
Q
explain Random forest
A
As the decision trees have a tendency towards overfitting, we try to fit N decision tree and then we assign the majority class during prediction
