Feature Crossing Hash

Problem

Hash a crossed categorical pair into a bucket. Build the key str(a) + '_' + str(b) and fold each character with h = (h * 31 + ord(ch)) mod num_buckets.

Return the bucket index; a non-positive num_buckets returns 0.

Hint

Crossing two categoricals then hashing keeps the feature space bounded.

Test cases

case 1

in: ["red","S",10]

expected: 3

case 2

in: [1,2,10]

expected: 4

case 3

in: ["cat","dog",7]

expected: 0

case 4

in: ["x","y",5]

expected: 1

Starter codePython

def feature_cross_hash(a, b, num_buckets):
    # Your code here
    pass

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