Random Feature Subset

Choose m distinct feature indices out of n_features, uniformly at random without replacement, using np.random.default_rng(seed).choice(n_features, size=m, replace=False). This is the feature subset a random-forest node considers when looking for a split.

Implement solve(n_features,m,seed=0).

Returns. Return an integer NumPy array of length m (in drawn order, not sorted). m must not exceed n_features.

Examples

Example 1

Input:

solve(5, 2, 0)

Output:

[3, 4]

Example 2

Input:

solve(10, 3, 7)

Output:

[7, 5, 6]