DTW-1NN Classification

Problem

Classify a query series with one-nearest-neighbor using dynamic time warping distance. Compute the minimum DTW distance from the query to every series in each class and return 0 or 1 for the closer class; class 0 wins ties.

An empty class has infinite distance. DTW uses absolute difference costs and the standard three-way recurrence.

Hint

DTW is a shape-aware distance, so scale differences still separate classes here.

Test cases

case 1

in: [[[1,2,3],[2,3,4]],[[10,11,12]],[1,2,4]]

expected: 0

case 2

in: [[[1,2,3],[2,3,4]],[[10,11,12]],[10,11,11]]

expected: 1

case 3

in: [[[1,1]],[[9,9]],[5,5]]

expected: 0

case 4

in: [[[0,0,0]],[[100,100,100]],[100,99,101]]

expected: 1

Starter codePython

def dtw_1nn_classify(train_class0, train_class1, query):
    # Your code here
    pass

4 test cases · run your code against them in the editor