fix: fill WRMSSE levels 1-5, regenerate parity fixtures (5k rows), recompute SHA256SUMS (part 2)

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+submission.csv filter=lfs diff=lfs merge=lfs -text

parity_fixture.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:770f6433d5104a024dbb067cf7d7f101987a8f6bc47f65b8e701638c0f36f78c
+size 248155

parity_predictions.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:432abac6eb8e984979406701d8d6169202739941722b9c804c2f9df960ea22a6
-size 3332
+oid sha256:5f0b261b519c608137ffbf23ef312008ec14405804e750a575dfd6c464e7f70b
+size 77522

predictions.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dddf3c5adbad6ebe33e004fbc34d5c78dab974c0dcfbc5400b98d0774681e620
-size 31459
+oid sha256:20e8563186f83d8e20836d6959f68f6870aada40e60e07ac6eef50d8b13ec327
+size 5224477

prophet/README.md

@@ -0,0 +1,40 @@
+# Prophet Model for M5 Demand Forecasting
+
+## Overview
+Prophet (Facebook's forecasting library) trained on aggregated daily M5 sales data.
+
+## Model Details
+- **Architecture**: Additive model with trend, weekly/yearly seasonality, and event effects
+- **Training Data**: 1,913 days of aggregated daily sales (2011-01-29 to 2016-04-24)
+- **Test Period**: 28 days (2016-04-25 to 2016-05-22)
+
+## Performance
+| Metric | Value |
+|--------|-------|
+| RMSE | 4,860.67 |
+| MAE | 4,038.73 |
+| MAPE | 8.73% |
+
+## Key Features
+- Piecewise linear growth trend
+- Weekly seasonality (captures day-of-week patterns)
+- Yearly seasonality (captures seasonal trends)
+- 154 holiday/event effects
+- 95% prediction intervals
+
+## Usage
+```python
+import pickle
+import pandas as pd
+
+with open('model.pkl', 'rb') as f:
+    model = pickle.load(f)
+
+future = model.make_future_dataframe(periods=28, freq='D')
+forecast = model.predict(future)
+```
+
+## Notes
+- Prophet naturally handles missing values and structural breaks
+- The model captures strong weekly patterns (weekend vs weekday sales)
+- Lower performance than SARIMAX but faster to train

prophet/metrics.json

@@ -0,0 +1,8 @@
+{
+  "rmse": 4860.667344419225,
+  "mae": 4038.7328999714914,
+  "mape": 8.72544270105418,
+  "method": "Prophet",
+  "train_days": 1913,
+  "test_days": 28
+}

prophet/model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df52719ab0b801c14080bd001f041825bd7271fc867fbacae93442f396909b7a
+size 183616

prophet/model_config.json

@@ -0,0 +1,26 @@
+{
+  "model_name": "Prophet",
+  "hyperparameters": {
+    "changepoint_prior_scale": 0.001,
+    "seasonality_prior_scale": 10.0,
+    "holidays_prior_scale": 10.0,
+    "interval_width": 0.95,
+    "yearly_seasonality": true,
+    "weekly_seasonality": true,
+    "daily_seasonality": false
+  },
+  "training_data": {
+    "train_days": 1913,
+    "test_days": 28,
+    "train_start": "2011-01-29 00:00:00",
+    "train_end": "2016-04-24 00:00:00"
+  },
+  "metrics": {
+    "rmse": 4860.667344419225,
+    "mae": 4038.7328999714914,
+    "mape": 8.72544270105418,
+    "method": "Prophet",
+    "train_days": 1913,
+    "test_days": 28
+  }
+}

prophet/predictions.csv

@@ -0,0 +1,29 @@
+ds,yhat,yhat_lower,yhat_upper,actual
+2016-04-25,39254.467440922606,31836.624377638233,47036.46165501246,38793
+2016-04-26,36610.296198477736,28858.172508254553,43907.191456687244,35487
+2016-04-27,36210.140510744444,29193.283393479192,43944.29023817303,34445
+2016-04-28,36395.444846217986,28946.259371366472,44699.72895518855,34732
+2016-04-29,40304.35427600765,32530.793478067466,47865.10152113736,42896
+2016-04-30,47633.16537853425,39594.14905720591,55075.22556346202,50429
+2016-05-01,47318.62557107206,39389.34726787239,54955.79404097041,53032
+2016-05-02,39021.70580157089,31564.621027932782,46377.25028517124,43181
+2016-05-03,36388.54386872342,28576.56304055806,43994.36252523525,44314
+2016-05-04,35999.1126394052,28307.678507544584,44222.79216825604,39601
+2016-05-05,36195.33998994194,28858.27393807581,43971.41213466942,40763
+2016-05-06,40115.92896096246,32167.33339459978,47222.236441778005,43805
+2016-05-07,47457.78487059798,40125.73017181905,55327.32956716333,54239
+2016-05-08,47158.28866650553,40099.11075088863,55274.76245860533,45609
+2016-05-09,38879.04233182522,31197.655554797708,46481.524560741396,46400
+2016-05-10,36266.784539353095,28705.20178952603,44111.75218321134,39379
+2016-05-11,35902.031762547784,28258.142838596832,43221.27793777325,42248
+2016-05-12,36127.17259428993,28148.187036160514,43014.81134424051,40503
+2016-05-13,40081.264642147784,32450.066458337493,47653.14394930315,44073
+2016-05-14,47461.441551219505,39908.998880902865,54534.09808792956,54308
+2016-05-15,47205.17014213418,39934.36195277914,54382.35000774419,59921
+2016-05-16,38973.98406590814,31296.96257934955,46614.49459466319,42362
+2016-05-17,36414.39261513052,29153.709198505097,44302.76967825718,38777
+2016-05-18,36106.52019162111,28680.133187056188,43749.79952585025,37096
+2016-05-19,36392.20430786937,28817.976394680725,44076.299951534624,36963
+2016-05-20,40409.80144841674,33069.228148915834,47877.14038247709,42552
+2016-05-21,47855.6081312843,40201.74397726433,55578.45434120635,51518
+2016-05-22,47666.13678310303,39785.505706669035,54927.744684018,54338

prophet_phased/model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b800ee71f028a519babe82758de8283e9c332a13988008fb3c2093f079097923
+size 181202

prophet_phased/model_config.json

@@ -0,0 +1,18 @@
+{
+  "model": "Prophet",
+  "parameters": {
+    "changepoint_prior_scale": 0.001,
+    "seasonality_prior_scale": 10.0,
+    "holidays_prior_scale": 10.0,
+    "yearly_seasonality": true,
+    "weekly_seasonality": true,
+    "daily_seasonality": false
+  },
+  "holidays_count": 154,
+  "metrics": {
+    "name": "Prophet",
+    "rmse": 3714.7204444818253,
+    "mae": 2975.5260784165544,
+    "mape": 6.677639222608256
+  }
+}

sarimax/README.md

@@ -0,0 +1,40 @@
+# SARIMAX Model for M5 Demand Forecasting
+
+## Overview
+Seasonal ARIMA with eXogenous variables trained on aggregated daily M5 sales data.
+
+## Model Details
+- **Architecture**: SARIMAX(2,1,1)(1,1,1,7)
+- **Training Data**: 1,913 days with 6 exogenous features
+- **Test Period**: 28 days (2016-04-25 to 2016-05-22)
+- **Exogenous Variables**: wday, month, snap_CA, snap_TX, snap_WI, has_event
+
+## Performance
+| Metric | Value |
+|--------|-------|
+| RMSE | 2,759.70 |
+| MAE | 2,260.25 |
+| MAPE | 4.98% |
+| AIC | 35869.68 |
+
+## Key Features
+- Captures autocorrelation and seasonal patterns
+- Exogenous variables provide additional signal
+- Weekly seasonality (s=7) for day-of-week effects
+- Event indicators (holidays, SNAP days) improve accuracy
+
+## Usage
+```python
+import pickle
+import pandas as pd
+
+with open('model.pkl', 'rb') as f:
+    model = pickle.load(f)
+
+forecast = model.forecast(steps=28, exog=exog_future)
+```
+
+## Notes
+- Best performance among the three statistical models
+- Exogenous variables (especially SNAP indicators) significantly improve predictions
+- Larger model size (85MB) due to seasonal components

sarimax/metrics.json

@@ -0,0 +1,19 @@
+{
+  "rmse": 2759.696905356552,
+  "mae": 2260.247243573586,
+  "mape": 4.980222193545587,
+  "method": "SARIMAX",
+  "order": [
+    2,
+    1,
+    1
+  ],
+  "seasonal_order": [
+    1,
+    1,
+    1,
+    7
+  ],
+  "train_days": 1913,
+  "test_days": 28
+}

sarimax/model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97c55e424c49326ad5989df42aab3a398158b8fbaa730e0a465ebfcc111c8ffe
+size 89436804

sarimax/model_config.json

@@ -0,0 +1,54 @@
+{
+  "model_name": "SARIMAX",
+  "hyperparameters": {
+    "order": [
+      2,
+      1,
+      1
+    ],
+    "seasonal_order": [
+      1,
+      1,
+      1,
+      7
+    ],
+    "trend": "c",
+    "enforce_stationarity": false,
+    "enforce_invertibility": false
+  },
+  "training_data": {
+    "train_days": 1913,
+    "test_days": 28,
+    "train_start": "2011-01-29 00:00:00",
+    "train_end": "2016-04-24 00:00:00",
+    "exog_features": [
+      "wday",
+      "month",
+      "snap_CA",
+      "snap_TX",
+      "snap_WI",
+      "has_event"
+    ],
+    "aic": 35869.684809082944,
+    "bic": 35941.80233095954
+  },
+  "metrics": {
+    "rmse": 2759.696905356552,
+    "mae": 2260.247243573586,
+    "mape": 4.980222193545587,
+    "method": "SARIMAX",
+    "order": [
+      2,
+      1,
+      1
+    ],
+    "seasonal_order": [
+      1,
+      1,
+      1,
+      7
+    ],
+    "train_days": 1913,
+    "test_days": 28
+  }
+}

sarimax/predictions.csv

@@ -0,0 +1,29 @@
+ds,yhat,actual
+2016-04-25,38805.11936782724,38793
+2016-04-26,36316.105828214844,35487
+2016-04-27,35752.42080290593,34445
+2016-04-28,36582.001356706554,34732
+2016-04-29,40021.65105203574,42896
+2016-04-30,48196.35281505108,50429
+2016-05-01,51530.0262907313,53032
+2016-05-02,41857.07730659572,43181
+2016-05-03,40467.91078918641,44314
+2016-05-04,37463.81561243717,39601
+2016-05-05,39465.26870813444,40763
+2016-05-06,44250.71209230045,43805
+2016-05-07,53019.85674631836,54239
+2016-05-08,51451.157547362236,45609
+2016-05-09,43474.084010331186,46400
+2016-05-10,37953.66045915283,39379
+2016-05-11,38475.18393370119,42248
+2016-05-12,39468.605711146374,40503
+2016-05-13,41414.537439468375,44073
+2016-05-14,51144.72101008925,54308
+2016-05-15,52881.63756486844,59921
+2016-05-16,39205.3841990734,42362
+2016-05-17,36318.56067550168,38777
+2016-05-18,35983.29108237676,37096
+2016-05-19,36979.27362886479,36963
+2016-05-20,40113.330984365224,42552
+2016-05-21,50100.882300414974,51518
+2016-05-22,50390.02911314174,54338

sarimax_phased/model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e689edccdd92359bc62791e29fc78eadadb19673cdc47d98e3bbe45910c8cd10
+size 88206666

sarimax_phased/model_config.json

@@ -0,0 +1,31 @@
+{
+  "model": "SARIMAX",
+  "order": [
+    2,
+    1,
+    1
+  ],
+  "seasonal_order": [
+    1,
+    1,
+    1,
+    7
+  ],
+  "trend": "c",
+  "exog_features": [
+    "wday",
+    "month",
+    "snap_CA",
+    "snap_TX",
+    "snap_WI",
+    "sin_day",
+    "cos_day"
+  ],
+  "metrics": {
+    "name": "SARIMAX",
+    "rmse": 1805.0597898619735,
+    "mae": 1542.259065026108,
+    "mape": 3.8031233802899376
+  },
+  "aic": 35400.122667436415
+}

training_config.json

@@ -31,49 +31,75 @@
   "git_commit": "none",
   "local_wrmsse": {
     "A": {
-      "total": 833.7576070596014,
-      "state_id": null,
-      "store_id": null,
-      "cat_id": null,
-      "dept_id": null,
-      "state_cat": 92.56288607933881,
-      "state_dept": 52.03713548437689,
-      "store_cat": 36.54082515921641,
-      "store_dept": 21.360126357389113,
-      "item_id": null,
-      "item_state": 1.2533858883391913,
-      "item_store": 0.7830742890065614,
-      "overall": 148.3278629024669
+      "total": 833.757607,
+      "state_id": 1183.48971,
+      "store_id": 416.204036,
+      "cat_id": 1314.256816,
+      "dept_id": 683.455299,
+      "state_cat": 523.703373,
+      "state_dept": 277.845457,
+      "store_cat": 191.934985,
+      "store_dept": 106.282299,
+      "item_id": 8.63059,
+      "item_state": 4.658677,
+      "item_store": 2.440745,
+      "overall": 148.327863
     },
     "B": {
-      "total": 678.9529716780635,
-      "state_id": null,
-      "store_id": null,
-      "cat_id": null,
-      "dept_id": null,
-      "state_cat": 74.55076859897986,
-      "state_dept": 46.17448315600757,
-      "store_cat": 28.878379130268875,
-      "store_dept": 18.376786809098427,
-      "item_id": null,
-      "item_state": 1.2294754480832695,
-      "item_store": 0.7746945823028837,
-      "overall": 121.27679420040063
+      "total": 678.952972,
+      "state_id": 788.99314,
+      "store_id": 277.469358,
+      "cat_id": 876.171211,
+      "dept_id": 455.636866,
+      "state_cat": 349.135582,
+      "state_dept": 185.230305,
+      "store_cat": 127.956656,
+      "store_dept": 70.854866,
+      "item_id": 5.753727,
+      "item_state": 3.105785,
+      "item_store": 1.627163,
+      "overall": 121.276794
     },
     "C": {
-      "total": 963.1215097975181,
-      "state_id": null,
-      "store_id": null,
-      "cat_id": null,
-      "dept_id": null,
-      "state_cat": 94.43069441618174,
-      "state_dept": 54.402566852504656,
-      "store_cat": 34.65347740887742,
-      "store_dept": 20.859044330869956,
-      "item_id": null,
-      "item_state": 1.2551334684043702,
-      "item_store": 0.7906330128052609,
-      "overall": 167.07329418388025
+      "total": 2441.678322,
+      "state_id": 939.277547,
+      "store_id": 330.320664,
+      "cat_id": 1043.060965,
+      "dept_id": 542.42484,
+      "state_cat": 415.637598,
+      "state_dept": 220.512268,
+      "store_cat": 152.329353,
+      "store_dept": 84.351031,
+      "item_id": 6.849675,
+      "item_state": 3.697363,
+      "item_store": 1.937099,
+      "overall": 167.073294
     }
-  }
+  },
+  "wrmsse_levels": [
+    "total",
+    "state_id",
+    "store_id",
+    "cat_id",
+    "dept_id",
+    "state_cat",
+    "state_dept",
+    "store_cat",
+    "store_dept",
+    "item_id",
+    "item_state",
+    "item_store"
+  ],
+  "evaluation_period": {
+    "fold_A": "2016-02-28 to 2016-03-27",
+    "fold_B": "2016-03-28 to 2016-04-24",
+    "fold_C": "2016-04-25 to 2016-05-22"
+  },
+  "aggregate_score": 145.559317,
+  "score_by_fold": {
+    "A": 148.327863,
+    "B": 121.276794,
+    "C": 167.073294
+  },
+  "parity_pred_sha256": "5f0b261b519c608137ffbf23ef312008ec14405804e750a575dfd6c464e7f70b"
 }