File size: 8,381 Bytes
976eb45 c059d87 976eb45 c059d87 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 | """
tests/test_crop_risk_scorer.py
================================
Validation tests for crop_risk_scorer.compute_risk_score.
These tests ensure:
- numerical stability
- monotonic risk behavior
- correct alert threshold transitions
- proper use of forecast signals
"""
import numpy as np
import pytest
from crop_risk_scorer import compute_risk_score, RiskWeights
from zone_observation import (
AlertLevel,
ForecastConfig,
make_synthetic_zone_obs,
make_synthetic_forecast_result,
)
# ---------------------------------------------------------------------------
# Fixtures
# ---------------------------------------------------------------------------
@pytest.fixture
def base_inputs():
obs = make_synthetic_zone_obs("z", seed=0)
fc = make_synthetic_forecast_result("z", horizon_days=14, valid_time=obs.valid_time)
return obs, fc
# ---------------------------------------------------------------------------
# Numerical sanity
# ---------------------------------------------------------------------------
class TestNumericalSanity:
def test_outputs_are_finite(self, base_inputs):
obs, fc = base_inputs
rs = compute_risk_score(obs, fc)
values = [
rs.supply_shortfall_prob,
rs.drought_risk,
rs.flood_risk,
rs.fungi_contamination_prob,
rs.quality_risk_composite,
rs.confidence,
]
for v in values:
assert np.isfinite(v), f"Non-finite value detected: {v}"
def test_outputs_in_unit_interval(self, base_inputs):
obs, fc = base_inputs
rs = compute_risk_score(obs, fc)
for field in [
rs.supply_shortfall_prob,
rs.drought_risk,
rs.flood_risk,
rs.fungi_contamination_prob,
rs.quality_risk_composite,
rs.confidence,
]:
assert 0.0 <= field <= 1.0
# ---------------------------------------------------------------------------
# Monotonicity (critical for RL learning)
# ---------------------------------------------------------------------------
class TestMonotonicity:
def test_drought_increase_raises_risk(self, base_inputs):
obs, fc = base_inputs
obs_low = obs
obs_high = make_synthetic_zone_obs("z", drought=True, seed=1)
rs_low = compute_risk_score(obs_low, fc)
rs_high = compute_risk_score(obs_high, fc)
assert rs_high.drought_risk >= rs_low.drought_risk
def test_flood_increase_raises_risk(self, base_inputs):
obs, fc = base_inputs
obs_low = obs
obs_high = make_synthetic_zone_obs("z", flood=True, seed=2)
rs_low = compute_risk_score(obs_low, fc)
rs_high = compute_risk_score(obs_high, fc)
assert rs_high.flood_risk >= rs_low.flood_risk
def test_combined_risk_raises_supply(self, base_inputs):
obs, fc = base_inputs
obs_low = obs
obs_high = make_synthetic_zone_obs("z", flood=True, drought=True, seed=3)
rs_low = compute_risk_score(obs_low, fc)
rs_high = compute_risk_score(obs_high, fc)
assert rs_high.supply_shortfall_prob >= rs_low.supply_shortfall_prob
# ---------------------------------------------------------------------------
# Forecast influence
# ---------------------------------------------------------------------------
class TestForecastInfluence:
def test_heavy_rain_forecast_increases_flood_risk(self, base_inputs):
obs, fc = base_inputs
fc_heavy = make_synthetic_forecast_result(
"z", horizon_days=14, valid_time=obs.valid_time, flood=True
)
rs_base = compute_risk_score(obs, fc)
rs_heavy = compute_risk_score(obs, fc_heavy)
assert rs_heavy.flood_risk >= rs_base.flood_risk
def test_drought_forecast_increases_drought_risk(self, base_inputs):
obs, fc = base_inputs
fc_dry = make_synthetic_forecast_result(
"z", horizon_days=14, valid_time=obs.valid_time, drought=True
)
rs_base = compute_risk_score(obs, fc)
rs_dry = compute_risk_score(obs, fc_dry)
assert rs_dry.drought_risk >= rs_base.drought_risk
# ---------------------------------------------------------------------------
# Alert thresholds (VERY important)
# ---------------------------------------------------------------------------
class TestAlertThresholds:
def test_critical_threshold(self, base_inputs):
obs, fc = base_inputs
from zone_observation import ForecastResult
obs_extreme = make_synthetic_zone_obs("z", drought=True, flood=True, seed=10)
obs_extreme.precip_anomaly_idx = -4.0 # maxes drought_signal()'s precip term
obs_extreme.soil_moisture_anom = -4.0 # maxes drought_signal()'s soil term
obs_extreme.flood_extent_pct = 100.0 # maxes flood_signal()'s extent term
obs_extreme.drainage_risk_idx = 1.0 # maxes flood_signal()'s drainage term
fc_extreme = ForecastResult(
zone_id=fc.zone_id,
forecast_time=fc.forecast_time,
horizon_days=fc.horizon_days,
precip_mm=fc.precip_mm,
precip_p10=fc.precip_p10,
precip_p90=fc.precip_p90,
temp_mean_c=fc.temp_mean_c,
temp_p10=fc.temp_p10,
temp_p90=fc.temp_p90,
rh_mean_pct=fc.rh_mean_pct,
prob_heavy_rain=tuple([1.0] * fc.horizon_days),
prob_drought_day=tuple([1.0] * fc.horizon_days),
prob_high_humidity=fc.prob_high_humidity,
source=fc.source,
)
rs = compute_risk_score(obs_extreme, fc_extreme)
assert rs.alert_level in [
AlertLevel.WARNING,
AlertLevel.CRITICAL,
]
def test_low_risk_produces_none_or_watch(self, base_inputs):
obs, fc = base_inputs
rs = compute_risk_score(obs, fc)
assert rs.alert_level in [
AlertLevel.NONE,
AlertLevel.WATCH,
AlertLevel.ADVISORY,
]
# ---------------------------------------------------------------------------
# Confidence model
# ---------------------------------------------------------------------------
class TestConfidence:
def test_confidence_in_unit_interval(self, base_inputs):
obs, fc = base_inputs
rs = compute_risk_score(obs, fc)
assert 0 <= rs.confidence <= 1
def test_observational_data_has_higher_confidence(self):
obs_obs = make_synthetic_zone_obs("z", seed=0)
fc_obs = make_synthetic_forecast_result("z", horizon_days=14, valid_time=obs_obs.valid_time)
obs_lowq = make_synthetic_zone_obs("z", seed=1)
obs_lowq.quality_flag = 3
rs_high = compute_risk_score(obs_obs, fc_obs)
rs_low = compute_risk_score(obs_lowq, fc_obs)
assert rs_high.confidence >= rs_low.confidence
# ---------------------------------------------------------------------------
# Stability / repeatability
# ---------------------------------------------------------------------------
class TestStability:
def test_same_inputs_same_output(self, base_inputs):
obs, fc = base_inputs
rs1 = compute_risk_score(obs, fc)
rs2 = compute_risk_score(obs, fc)
assert rs1.supply_shortfall_prob == pytest.approx(rs2.supply_shortfall_prob)
def test_no_nan_under_extreme_inputs(self):
obs = make_synthetic_zone_obs("z", seed=99)
fc_base = make_synthetic_forecast_result("z", horizon_days=30, valid_time=obs.valid_time)
from zone_observation import ForecastResult
fc = ForecastResult(
zone_id=fc_base.zone_id,
forecast_time=fc_base.forecast_time,
horizon_days=30,
precip_mm=tuple([500.0] * 30),
precip_p10=tuple([400.0] * 30),
precip_p90=tuple([500.0] * 30),
temp_mean_c=fc_base.temp_mean_c,
temp_p10=fc_base.temp_p10,
temp_p90=fc_base.temp_p90,
rh_mean_pct=fc_base.rh_mean_pct,
prob_heavy_rain=tuple([1.0] * 30),
prob_drought_day=tuple([0.0] * 30),
prob_high_humidity=fc_base.prob_high_humidity,
source=fc_base.source,
)
rs = compute_risk_score(obs, fc)
assert np.isfinite(rs.supply_shortfall_prob) |