# Generate a random index for train/test split
num_rows = df.shape[0]
frac_test = 0.1
num_test_samples = int(num_rows * frac_test)

idx = np.full(num_rows, False)
idx[:num_test_samples] = True
np.random.seed(8)
np.random.shuffle(idx)

print(idx.shape)
idx