band.py

import numpy as np

# Band data set
# A test data set meant to an imbalanced data set with a small positive class (4.886%)
# Randomly sample [0, 1) x [0, 1)
# Points within the annulus with
#  - c = [0.5, 0.5]
#  - 0.38 < r < 0.4
# Are Positive (1)
# All other points are Negative (0)
np.random.seed(seed=0)
def make_dataset(n):
    x = np.random.rand(n, 2)
    y = []
    for i in range(0, n):
        x_i = x[i]
        r = np.linalg.norm(x_i - 0.5)
        if r < 0.4 and r > 0.38:
            y.append(1)
        else:
            y.append(0)
            
    return (x, y)

def test_dataset():
    avg = 0
    runs = 100
    n = 5000
    for i in range(0, runs):
        (_, y) = make_dataset(n)
        run_ratio = float(np.sum(y)) / n
        avg += run_ratio / runs
    print(f"band test, avg positive percentage: {avg * 100}")