Thibaut Forest Claude Opus 5 (1M context) commited on
Commit ·
9d114fd
1
Parent(s): 24b7f13
CV v4 (user directive): building-GroupKFold x temporal purge - per fold, per scenario, fit on train buildings' pairs with label windows entirely outside the scenario window; test scenario on the test buildings as a standalone fleet
Browse filesDouble holdout by construction (fleet axis + time axis), no tuning: the
public-calibrated pricing layer was explicitly rejected. Coherence with
recorded public deltas is checked, never fitted.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- batteryswap_example/cv.py +41 -27
batteryswap_example/cv.py
CHANGED
|
@@ -1,13 +1,20 @@
|
|
| 1 |
-
"""THE CV (user directive 2026-08-22
|
| 2 |
|
| 3 |
ONE protocol lives here, deliberately. Buildings are assigned to K folds,
|
| 4 |
stratified on battery-failure count and location (travel centrality). For
|
| 5 |
-
each fold
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
THE SCORE is the mean over folds of the fold's mean scenario cost.
|
| 10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
Why this protocol (2026-08-22, the day the temporal purge fell):
|
| 12 |
|
| 13 |
* It simulates the HELD-OUT deployment path: the competition's private split
|
|
@@ -20,14 +27,16 @@ Why this protocol (2026-08-22, the day the temporal purge fell):
|
|
| 20 |
count bias (phase 26): the count rule operates on the test fleet's own
|
| 21 |
sum(p42), exactly as deployment does on a held-out fleet.
|
| 22 |
|
| 23 |
-
*
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
|
|
|
|
|
|
| 31 |
|
| 32 |
* Diagnostics are PLAN-DERIVED only (cost blocks + in-window swap count), so
|
| 33 |
ANY method that yields a plan can be scored. Candidates must win at ~flat
|
|
@@ -151,13 +160,14 @@ VARIANTS = {'base': _base, 'ranks_add': _ranks_add, 'vrel': _vrel}
|
|
| 151 |
|
| 152 |
def run_bgkf(dataset_path=Path('dataset'), kind=None, variant='base',
|
| 153 |
blend=None, n_folds=N_FOLDS, origins=range(48)):
|
| 154 |
-
"""THE SCORE. For each fold
|
| 155 |
-
|
| 156 |
-
standalone fleet and evaluate. Floor off
|
| 157 |
-
deployed floor is a submit.py concern).
|
| 158 |
-
model family (mc.MODELS key or
|
| 159 |
-
set, `blend`=(kind2, w2) a
|
| 160 |
-
|
|
|
|
| 161 |
"""
|
| 162 |
kind = kind or CHAMPION['model']
|
| 163 |
t = load_table(dataset_path)
|
|
@@ -188,12 +198,16 @@ def run_bgkf(dataset_path=Path('dataset'), kind=None, variant='base',
|
|
| 188 |
recs = []
|
| 189 |
for f in range(n_folds):
|
| 190 |
test_b = set(folds.index[folds == f])
|
| 191 |
-
|
| 192 |
-
model = _fit(kind, train)
|
| 193 |
-
partner = _fit(blend[0], train) if blend is not None else None
|
| 194 |
for idx in origins:
|
| 195 |
c = ctx[idx]
|
| 196 |
T = c['start_time']
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 197 |
fleet = c['locs'][c['locs']['building'].isin(test_b)]
|
| 198 |
rows = (t[t['origin_idx'] == idx].set_index('device_id')
|
| 199 |
.reindex(fleet['battery']).dropna(subset=['v_last']))
|
|
@@ -312,10 +326,10 @@ def main():
|
|
| 312 |
candidate_report()
|
| 313 |
else:
|
| 314 |
raise SystemExit(
|
| 315 |
-
f'unknown protocol {proto!r} - the building-GroupKFold
|
| 316 |
-
f'is the only CV (user directive 2026-08-22
|
| 317 |
-
f'purge lives in git at f5f3fda; the
|
| 318 |
-
f'at 2000281.')
|
| 319 |
print(f'\n[{time.time() - t0:.0f}s]')
|
| 320 |
|
| 321 |
|
|
|
|
| 1 |
+
"""THE CV (user directive 2026-08-22 v4): building-GroupKFold x temporal purge.
|
| 2 |
|
| 3 |
ONE protocol lives here, deliberately. Buildings are assigned to K folds,
|
| 4 |
stratified on battery-failure count and location (travel centrality). For
|
| 5 |
+
each fold, for each of the 48 scenario times: fit on the TRAIN buildings'
|
| 6 |
+
pairs whose 42d label windows lie ENTIRELY outside the scenario window
|
| 7 |
+
(origin + W <= T or origin >= T + W - the label-overlap purge), then plan
|
| 8 |
+
the TEST buildings as a STANDALONE fleet and evaluate its cost.
|
| 9 |
THE SCORE is the mean over folds of the fold's mean scenario cost.
|
| 10 |
|
| 11 |
+
Double holdout, no tuning: the fit never sees the test buildings (fleet
|
| 12 |
+
axis) AND never sees the scenario's own window (time axis). Coherence with
|
| 13 |
+
the public board is CHECKED against the recorded public deltas, never
|
| 14 |
+
fitted - a calibrated pricing layer was explicitly rejected (user,
|
| 15 |
+
2026-08-22): an instrument tuned to reproduce past public deltas can only
|
| 16 |
+
confirm what it was tuned to.
|
| 17 |
+
|
| 18 |
Why this protocol (2026-08-22, the day the temporal purge fell):
|
| 19 |
|
| 20 |
* It simulates the HELD-OUT deployment path: the competition's private split
|
|
|
|
| 27 |
count bias (phase 26): the count rule operates on the test fleet's own
|
| 28 |
sum(p42), exactly as deployment does on a held-out fleet.
|
| 29 |
|
| 30 |
+
* Instrument history: the pure temporal purge (fleet-wide) fell to three
|
| 31 |
+
public refutations in one day (phase 56: count-coupling, label-process
|
| 32 |
+
ghosts, era-calibration leak - its fits saw the future of the same fleet).
|
| 33 |
+
The pure building split (v3) fell the same evening: it still sign-flipped
|
| 34 |
+
all three public killers (phase 57 alignment check) because its fits saw
|
| 35 |
+
the scenario's own era through other buildings. v4 removes both channels
|
| 36 |
+
structurally. What NO local split can price (public-measured, 4x): the
|
| 37 |
+
public fleet's uncatchable tail and its train-absent ghost devices -
|
| 38 |
+
count-coupled and label-process candidates stay report-only regardless of
|
| 39 |
+
what this harness says.
|
| 40 |
|
| 41 |
* Diagnostics are PLAN-DERIVED only (cost blocks + in-window swap count), so
|
| 42 |
ANY method that yields a plan can be scored. Candidates must win at ~flat
|
|
|
|
| 160 |
|
| 161 |
def run_bgkf(dataset_path=Path('dataset'), kind=None, variant='base',
|
| 162 |
blend=None, n_folds=N_FOLDS, origins=range(48)):
|
| 163 |
+
"""THE SCORE. For each fold, per scenario: fit on the TRAIN buildings'
|
| 164 |
+
purged pairs (label windows fully outside the scenario window), then
|
| 165 |
+
plan the TEST buildings as a standalone fleet and evaluate. Floor off
|
| 166 |
+
(harness minimalism; the deployed floor is a submit.py concern).
|
| 167 |
+
Candidate hooks: `kind` swaps the model family (mc.MODELS key or
|
| 168 |
+
SklearnRUL spec), `variant` the feature set, `blend`=(kind2, w2) a
|
| 169 |
+
value-preserving rank blend (order changes, the primary's p42 values
|
| 170 |
+
and hence the count machinery do not).
|
| 171 |
"""
|
| 172 |
kind = kind or CHAMPION['model']
|
| 173 |
t = load_table(dataset_path)
|
|
|
|
| 198 |
recs = []
|
| 199 |
for f in range(n_folds):
|
| 200 |
test_b = set(folds.index[folds == f])
|
| 201 |
+
fold_train = closed[~t_bldg.loc[closed.index].isin(test_b)]
|
|
|
|
|
|
|
| 202 |
for idx in origins:
|
| 203 |
c = ctx[idx]
|
| 204 |
T = c['start_time']
|
| 205 |
+
# the purge, inside the fold: train pairs' label windows must lie
|
| 206 |
+
# entirely before or after the scenario window
|
| 207 |
+
train = fold_train[(fold_train['origin'] + W <= T)
|
| 208 |
+
| (fold_train['origin'] >= T + W)]
|
| 209 |
+
model = _fit(kind, train)
|
| 210 |
+
partner = _fit(blend[0], train) if blend is not None else None
|
| 211 |
fleet = c['locs'][c['locs']['building'].isin(test_b)]
|
| 212 |
rows = (t[t['origin_idx'] == idx].set_index('device_id')
|
| 213 |
.reindex(fleet['battery']).dropna(subset=['v_last']))
|
|
|
|
| 326 |
candidate_report()
|
| 327 |
else:
|
| 328 |
raise SystemExit(
|
| 329 |
+
f'unknown protocol {proto!r} - the building-GroupKFold x purge '
|
| 330 |
+
f'harness is the only CV (user directive 2026-08-22 v4). The '
|
| 331 |
+
f'fleet-wide purge lives in git at f5f3fda; the pure building '
|
| 332 |
+
f'split at 4be2283; the walk-forward family at 2000281.')
|
| 333 |
print(f'\n[{time.time() - t0:.0f}s]')
|
| 334 |
|
| 335 |
|