78 lines
2.5 KiB
Python
78 lines
2.5 KiB
Python
import math
|
|
import os
|
|
import sys
|
|
|
|
import pytest
|
|
|
|
# Ensure project root is importable when tests run in isolated environments.
|
|
sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
|
|
|
|
from app.services.normalization import apply_normalization
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
("nm", "rataan", "sb", "expected"),
|
|
[
|
|
(500, 500, 100, 500),
|
|
(600, 500, 100, 600),
|
|
(400, 500, 100, 400),
|
|
(1000, 500, 100, 1000),
|
|
(0, 500, 100, 0),
|
|
(500, 600, 80, 375),
|
|
],
|
|
)
|
|
def test_apply_normalization_nominal_cases(nm: int, rataan: float, sb: float, expected: int):
|
|
assert apply_normalization(nm, rataan, sb) == expected
|
|
|
|
|
|
@pytest.mark.parametrize("nm", [-1, 1001, 1500, -100])
|
|
def test_apply_normalization_rejects_invalid_nm(nm: int):
|
|
with pytest.raises(ValueError):
|
|
apply_normalization(nm, 500, 100)
|
|
|
|
|
|
@pytest.mark.parametrize("sb", [0, -1, -100.0])
|
|
def test_apply_normalization_returns_default_when_sd_non_positive(sb: float):
|
|
assert apply_normalization(500, 500, sb) == 500
|
|
|
|
|
|
def test_dynamic_normalization_distribution_behaves_as_expected():
|
|
nm_scores = [450, 480, 500, 520, 550, 480, 510, 490, 530, 470]
|
|
|
|
mean = sum(nm_scores) / len(nm_scores)
|
|
variance = sum((x - mean) ** 2 for x in nm_scores) / len(nm_scores)
|
|
std = math.sqrt(variance)
|
|
|
|
nn_scores = [apply_normalization(nm, mean, std) for nm in nm_scores]
|
|
nn_mean = sum(nn_scores) / len(nn_scores)
|
|
nn_variance = sum((x - nn_mean) ** 2 for x in nn_scores) / len(nn_scores)
|
|
nn_std = math.sqrt(nn_variance)
|
|
|
|
# Rounding in apply_normalization introduces small drift; these bounds are tight.
|
|
assert abs(nn_mean - 500) <= 5
|
|
assert abs(nn_std - 100) <= 5
|
|
|
|
|
|
def test_incremental_population_stats_match_batch_stats():
|
|
scores = [500, 550, 450, 600, 400]
|
|
|
|
participant_count = 0
|
|
total_nm_sum = 0.0
|
|
total_nm_sq_sum = 0.0
|
|
|
|
for score in scores:
|
|
participant_count += 1
|
|
total_nm_sum += score
|
|
total_nm_sq_sum += score * score
|
|
|
|
incremental_mean = total_nm_sum / participant_count
|
|
incremental_variance = (total_nm_sq_sum / participant_count) - (incremental_mean**2)
|
|
incremental_std = math.sqrt(max(0.0, incremental_variance))
|
|
|
|
batch_mean = sum(scores) / len(scores)
|
|
batch_variance = sum((x - batch_mean) ** 2 for x in scores) / len(scores)
|
|
batch_std = math.sqrt(batch_variance)
|
|
|
|
assert incremental_mean == pytest.approx(batch_mean, rel=0, abs=1e-10)
|
|
assert incremental_std == pytest.approx(batch_std, rel=0, abs=1e-10)
|