File size: 10,331 Bytes
bf3cbf1 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 | """Integration tests for real functionality: persistence, CSV import, and statistical attribution."""
from __future__ import annotations
import os
import tempfile
from io import StringIO
from pathlib import Path
import pytest
from spinor_os import ExperimentationOS, PersistenceManager, import_events_csv
from spinor_os.config import AttributionMethod, EventType
def test_csv_import_allocates_missions_and_records_events():
os = ExperimentationOS()
os.register_employee("import-001", "field-scientist", "northeast")
hypothesis = os.propose_hypothesis(
statement="Imported outreach increases appointments.",
causal_claim="Tailored outreach causes higher appointment rate.",
predicted_effect={
"metric": "appointment_rate",
"direction": "increase",
"magnitude": 0.1,
"unit": "percentage_point",
"timing": "7d",
},
employee_owner="import-001",
falsification_criteria=["No increase observed."],
customer_segment="enterprise",
territory="northeast",
modification="imported_outreach",
)
experiment = os.start_experiment(hypothesis.hypothesis_id)
csv_text = f"""experiment_id,actor_id,event_type,outcome_value,metric,execution_quality,actor_role
{experiment.experiment_id},import-001,OUTCOME_OBSERVED,0.15,appointment_rate,0.88,field-scientist
{experiment.experiment_id},import-001,OUTCOME_OBSERVED,0.18,appointment_rate,0.90,field-scientist
"""
summary = import_events_csv(os, StringIO(csv_text), experiment_id=experiment.experiment_id)
assert summary["rows_read"] == 2
assert summary["rows_ingested"] == 2
assert len(summary["event_ids"]) == 2
assert not summary["errors"]
assert len(os.events) == 2
def test_real_attribution_ols_and_diff_in_diff():
os = ExperimentationOS()
os.register_employee("alice", "field-scientist", "northeast")
os.register_employee("bob", "field-scientist", "southwest")
hypothesis = os.propose_hypothesis(
statement="X lifts Y in enterprise segment",
causal_claim="X causes Y",
predicted_effect={
"metric": "y_rate",
"direction": "increase",
"magnitude": 0.1,
"unit": "pp",
"timing": "7d",
},
employee_owner="alice",
falsification_criteria=["no effect"],
customer_segment="enterprise",
territory="northeast",
modification="treatment_a",
)
treat = os.start_experiment(hypothesis.hypothesis_id)
control = os.start_experiment(hypothesis.hypothesis_id)
m_t = os.allocate_mission(
treat.experiment_id,
"alice",
[
{
"hypothesis_id": hypothesis.hypothesis_id,
"modification": "treatment_a",
"customer_segment": "enterprise",
"territory": "northeast",
"timing": "2026-W31",
"resource_allocation": 1.0,
}
],
)
m_c = os.allocate_mission(
control.experiment_id,
"bob",
[
{
"hypothesis_id": hypothesis.hypothesis_id,
"modification": "control",
"customer_segment": "enterprise",
"territory": "southwest",
"timing": "2026-W31",
"resource_allocation": 1.0,
}
],
)
os.record_event(
treat.experiment_id,
m_t.mission_id,
EventType.OUTCOME_OBSERVED,
"alice",
outcome_value=0.05,
metric="y_rate",
execution_quality=0.9,
)
os.record_event(
control.experiment_id,
m_c.mission_id,
EventType.OUTCOME_OBSERVED,
"bob",
outcome_value=0.07,
metric="y_rate",
execution_quality=0.9,
)
os.record_event(
treat.experiment_id,
m_t.mission_id,
EventType.OUTCOME_OBSERVED,
"alice",
outcome_value=0.35,
metric="y_rate",
execution_quality=0.9,
)
os.record_event(
control.experiment_id,
m_c.mission_id,
EventType.OUTCOME_OBSERVED,
"bob",
outcome_value=0.07,
metric="y_rate",
execution_quality=0.85,
)
claim_ols = os.attribute(
treat.experiment_id,
"y_rate",
0.35,
0.05,
AttributionMethod.OLS,
0.95,
falsification_survived=True,
)
assert claim_ols.real_attribution is not None
assert claim_ols.real_attribution["method"] == "ols_regression"
assert claim_ols.real_attribution["estimated_effect"] > 0
assert claim_ols.real_attribution["n_observations"] >= 4
claim_did = os.attribute(
treat.experiment_id,
"y_rate",
0.35,
0.05,
AttributionMethod.DIFF_IN_DIFF,
0.95,
falsification_survived=True,
)
assert claim_did.real_attribution is not None
assert claim_did.real_attribution["method"] == "difference_in_differences"
assert claim_did.real_attribution["estimated_effect"] > 0
assert claim_did.real_attribution["n_observations"] >= 4
def test_persistence_reloads_imported_data():
with tempfile.TemporaryDirectory() as tmp:
db = Path(tmp) / "spinor_real.sqlite"
os1 = ExperimentationOS(persistence=PersistenceManager(db))
os1.register_employee("persist-001", "field-scientist", "northeast")
hypothesis = os1.propose_hypothesis(
statement="Persistent outreach works.",
causal_claim="Outreach causes lift.",
predicted_effect={
"metric": "lift",
"direction": "increase",
"magnitude": 0.1,
"unit": "pp",
"timing": "7d",
},
employee_owner="persist-001",
falsification_criteria=["no lift"],
customer_segment="enterprise",
territory="northeast",
modification="outreach",
)
experiment = os1.start_experiment(hypothesis.hypothesis_id)
csv_text = f"""experiment_id,actor_id,event_type,outcome_value,metric,execution_quality
{experiment.experiment_id},persist-001,OUTCOME_OBSERVED,0.12,lift,0.88
"""
summary = import_events_csv(os1, StringIO(csv_text), experiment_id=experiment.experiment_id)
assert summary["rows_ingested"] == 1
# Reopen from the same SQLite file.
os2 = ExperimentationOS(persistence=PersistenceManager(db))
assert "persist-001" in os2.employees
assert experiment.experiment_id in os2.experiments
assert len(os2.events) == 1
def test_server_import_and_real_attribution(client):
"""Live-server style verification using the FastAPI TestClient."""
# Register employee and hypothesis.
r = client.post("/employees", json={
"employee_id": "real-api-001",
"role": "field-scientist",
"territory": "northeast",
})
assert r.status_code == 200
r = client.post("/hypotheses", json={
"statement": "Real outreach increases conversions",
"causal_claim": "Outreach causes conversions",
"predicted_effect": {
"metric": "real_conversion_rate",
"direction": "increase",
"magnitude": 0.1,
"unit": "pp",
"timing": "7d",
},
"employee_owner": "real-api-001",
"falsification_criteria": ["No lift"],
"customer_segment": "enterprise",
"territory": "northeast",
"modification": "real_outreach",
})
assert r.status_code == 200
hypothesis_id = r.json()["hypothesis_id"]
# Create treatment and control experiments.
treat = client.post("/experiments", json={"hypothesis_id": hypothesis_id}).json()
control = client.post("/experiments", json={"hypothesis_id": hypothesis_id}).json()
treat_id = treat["experiment_id"]
control_id = control["experiment_id"]
# Allocate missions.
m_t = client.post(f"/experiments/{treat_id}/missions", json={
"employee_id": "real-api-001",
"candidates": [{
"hypothesis_id": hypothesis_id,
"modification": "real_outreach",
"customer_segment": "enterprise",
"territory": "northeast",
"timing": "2026-W31",
"resource_allocation": 1.0,
}],
}).json()
m_c = client.post(f"/experiments/{control_id}/missions", json={
"employee_id": "real-api-001",
"candidates": [{
"hypothesis_id": hypothesis_id,
"modification": "control",
"customer_segment": "enterprise",
"territory": "northeast",
"timing": "2026-W31",
"resource_allocation": 1.0,
}],
}).json()
# Record pre and post outcomes.
for (exp, mid, val, eq) in [
(treat_id, m_t["mission_id"], 0.05, 0.9),
(control_id, m_c["mission_id"], 0.06, 0.9),
(treat_id, m_t["mission_id"], 0.35, 0.9),
(control_id, m_c["mission_id"], 0.07, 0.85),
]:
r = client.post(f"/experiments/{exp}/events", json={
"mission_id": mid,
"event_type": "outcome_observed",
"actor_id": "real-api-001",
"outcome_value": val,
"metric": "real_conversion_rate",
"execution_quality": eq,
})
assert r.status_code == 200
# Compute real OLS attribution.
r = client.post(f"/experiments/{treat_id}/attributes", json={
"outcome_metric": "real_conversion_rate",
"outcome_value": 0.35,
"counterfactual_estimate": 0.05,
"method": "ols",
"confidence": 0.95,
"falsification_survived": True,
})
assert r.status_code == 200
data = r.json()
assert data["real_attribution"] is not None
assert data["real_attribution"]["method"] == "ols_regression"
assert data["real_attribution"]["estimated_effect"] > 0
# Force a save and check persistence status.
r = client.post("/persistence/save")
assert r.status_code == 200
r = client.get("/persistence/status")
assert r.status_code == 200
status = r.json()
assert status["persistence_enabled"] is True
assert status["experiments"] >= 2
|