ordersfev / AutoGluon ↗Rohlik E-Grocery Daily Fulfillment Orders
Daily order fulfillment volumes across 7 metropolitan fulfillment centers operated by European e-grocery leader Rohlik Group, covering a 61-day forward forecast horizon. The primary objective is multi-site demand forecasting utilizing calendar events, local weather, and web telemetry to optimize warehouse shift capacity and supply chain replenishment.
Business Objective: Rohlik E-Grocery Daily Fulfillment Orders
01Business Context
Rohlik operates rapid-delivery fulfillment centers across European metropolitan areas (including Prague, Brno, Munich, and Frankfurt). Warehouse managers and operations planners must schedule picker shifts, delivery driver fleets, and perishable stock procurement 1 to 61 days ahead to handle daily fluctuating grocery demand driven by weather and local calendar anomalies.
02Analytical Objective
Predict daily fulfillment order counts (`orders`) for each warehouse location over a 61-day horizon H, accounting for known public holidays, school recesses, brick-and-mortar retail closures, and historical web telemetry.
03Economic & Decision Impact
Under-forecasting leads to delivery slot throttling, severe last-mile delivery delays, and lost gross merchandise value (GMV). Over-forecasting creates excess labor costs from idle warehouse pickers and couriers, as well as acute food spoilage across short-shelf-life fresh grocery categories.
Classical univariate methods (AutoARIMA, ETS) capture baseline weekly seasonality but struggle to incorporate disparate holiday calendars across multiple European countries and sudden weather shifts. Gradient boosted trees with recursive lag features (LightGBM, CatBoost) achieve competitive baselines when engineered with regional calendar encodings and promotional interaction terms. Zero-shot foundation models, such as Mistral's Timestral and MOIRAI, demonstrate strong zero-shot generalization across multi-center e-grocery demand curves by directly ingesting holiday covariates and recent web engagement telemetry in-context without retraining.
Target & Feature Column Definitions
Exhaustive business definitions, measurement units, roles, and target variables across all 17 columns.
| Col | Variable / Feature Name | Role | Type | Unit / Scale | Description & Business Meaning |
|---|---|---|---|---|---|
| C | ordersFulfilled Orders | TARGET | NUM | count | Total number of completed grocery fulfillment orders delivered on the given date (primary target variable) |
| A | idFulfillment Center ID | SERIES ID | CAT | — | Unique identifier for the metropolitan fulfillment warehouse center (e.g. Brno_1, Prague_1, Munich_1) |
| B | timestampOrder Date | TIME | TIME | — | Calendar observation date formatted at daily granularity (YYYY-MM-DD) |
| D | holidayPublic Holiday Flag | Feature | NUM | — | Binary indicator specifying whether the date is a recognized public or bank holiday in the warehouse region |
| E | school_holidaysSchool Holiday Flag | Feature | NUM | — | Binary indicator representing active primary and secondary school vacation periods in the regional market |
| F | shops_closedShops Closed Flag | Feature | NUM | — | Binary indicator reflecting mandatory physical retail store closure mandates (e.g., German Sunday trading bans or statutory holiday closures), driving shopping volume online |
| G | winter_school_holidaysWinter School Holiday Flag | Feature | NUM | — | Binary flag denoting regional mid-winter and spring break school holiday windows |
| H | blackoutBlackout Event | Feature | NUM | — | Binary flag indicating an unscheduled facility power outage or major IT infrastructure disruption |
| I | frankfurt_shutdownFrankfurt Shutdown Flag | Feature | NUM | — | Binary indicator of localized operational stoppage or warehouse commissioning downtime at the Frankfurt distribution center |
| J | mini_shutdownPartial Shutdown Flag | Feature | NUM | — | Binary flag indicating localized equipment, automation conveyor maintenance, or partial intra-day dispatch delays |
| K | mov_changeMinimum Order Value Policy Shift | Feature | NUM | — | Indicator of revisions to minimum order value thresholds or delivery fee structure influencing basket sizing and transaction frequency |
| L | precipitationPrecipitation | Feature | NUM | mm | Recorded rainfall or liquid equivalent precipitation depth in the metropolitan service area |
| M | shutdownFacility Full Shutdown | Feature | NUM | — | Binary flag indicating a complete suspension of warehouse order fulfillment operations |
| N | snowSnowfall Depth | Feature | NUM | cm | Measured snowfall accumulation impacting courier travel speeds and customer stay-at-home ordering behavior |
| O | user_activity_1Digital Channel Traffic Index (Primary) | Feature | NUM | count | Aggregated upstream digital search, app engagement, or active shopper session telemetry preceding order checkout |
| P | user_activity_2Digital Channel Traffic Index (Secondary) | Feature | NUM | count | High-volume web interaction and catalog browsing activity metric capturing real-time shopping intent |
| Q | holiday_nameHoliday Name | Feature | CAT | — | Text descriptor of specific cultural, religious, or statutory holiday events (e.g. Christmas Eve, Easter Monday, New Year's Day) |
Interactive Data Table
Explore rows, feature values, and target labels for Rohlik E-Grocery Daily Fulfillment Orders.
Foundation Models vs Traditional ML & Baselines (41 Methods)
Empirical evaluation directly from the AutoGluon fev benchmark suite evaluating zero-shot foundation models against deep learning, gradient-boosted trees, and statistical baselines across 100 evaluation tasks.
| Rank | Method / Model | Family | Organization | SQL Win Rate?Average win rate (%) against all competing models across the benchmark tasks under Scaled Quantile Loss. | SQL Skill?Normalized skill score relative to the Seasonal Naive baseline (0% = baseline, higher is better). | MASE Win Rate?Average win rate (%) under Mean Absolute Scaled Error. | Latency?Median end-to-end execution time per 100 time series in seconds. | Link |
|---|---|---|---|---|---|---|---|---|
| #1 | Timestral-209M | Foundation Model | Mistral AI | 87.7% | +50.4% | 86.0% | 0.51s | Repo ↗ |
| #2 | Timestral-123M (Carla Engine) | Foundation Model | Mistral AI | 86.8% | +50.2% | 85.0% | 0.32s | Repo ↗ |
| #3 | TimeRSI | Closed-Source API | GaTech & SamayLabs | 84.8% | +51.2% | 83.8% | 2512s | Repo ↗ |
| #4 | t0-gamma | Foundation Model | The Forecasting Company | 84.4% | +49.4% | 81.8% | 4.38s | Repo ↗ |
| #5 | Timestral-28M | Foundation Model | Mistral AI | 79.2% | +48.4% | 77.0% | 0.11s | Repo ↗ |
| #6 | timesfm-3 | Foundation Model | 78.5% | +49.0% | 77.0% | 3.41s | Repo ↗ | |
| #7 | ChakraTS | Baseline | Community | 78.2% | +48.1% | 79.5% | 14.22s | Repo ↗ |
| #8 | Lingjiang2.0 | Baseline | Community | 73.9% | +47.0% | 71.3% | 8.89s | Repo ↗ |
| #9 | Chronos-2 | Foundation Model | AWS | 70.4% | +47.3% | 68.4% | 0.84s | Repo ↗ |
| #10 | timeh2o-v1-tiny | Foundation Model | H2O.ai | 69.9% | +45.5% | 67.0% | 0.19s | Repo ↗ |
| #11 | Timestral-5M | Foundation Model | Mistral AI | 69.0% | +46.2% | 67.7% | 40ms | Repo ↗ |
| #12 | Toto-2.0-2.5B | Baseline | Community | 67.5% | +44.4% | 65.7% | 5.58s | Repo ↗ |
| #13 | t0-beta | Foundation Model | The Forecasting Company | 67.4% | +46.6% | 64.5% | 2.29s | Repo ↗ |
| #14 | Toto-2.0-1B | Baseline | Community | 66.4% | +44.4% | 65.5% | 2.48s | Repo ↗ |
| #15 | TiRex-2 | Foundation Model | NX-AI | 65.8% | +45.5% | 63.4% | 0.27s | Repo ↗ |
| #16 | Toto-2.0-313m | Baseline | Community | 65.1% | +44.2% | 63.5% | 1.06s | Repo ↗ |
| #17 | TiRex | Foundation Model | NX-AI | 58.0% | +42.6% | 54.1% | 0.24s | Repo ↗ |
| #18 | Toto-2.0-22m | Baseline | Community | 57.6% | +42.9% | 56.0% | 0.25s | Repo ↗ |
| #19 | TimesFM-2.5 | Baseline | Community | 54.6% | +42.2% | 53.5% | 1.87s | Repo ↗ |
| #20 | TabPFN-TS-3 | Foundation Model | Prior Labs | 52.4% | +43.1% | 51.3% | 235s | Repo ↗ |
| #21 | TS-ICL | Foundation Model | EDF-Lab | 51.2% | +43.1% | 49.8% | 2.49s | Repo ↗ |
| #22 | FlowState | Foundation Model | IBM | 49.3% | +39.9% | 49.9% | 2.29s | Repo ↗ |
| #23 | citras-fm | Foundation Model | Hitachi | 48.8% | +41.2% | 46.9% | 0.60s | Repo ↗ |
| #24 | Toto-1.0 | Baseline | Community | 47.5% | +40.8% | 45.9% | 22.35s | Repo ↗ |
| #25 | TabPFN-TS | Foundation Model | Prior Labs | 47.3% | +41.5% | 45.4% | 109s | Repo ↗ |
| #26 | Toto-2.0-4m | Baseline | Community | 46.2% | +40.7% | 43.6% | 0.20s | Repo ↗ |
| #27 | Moirai-2.0 | Baseline | Community | 41.9% | +39.3% | 40.5% | 0.35s | Repo ↗ |
| #28 | Chronos-Bolt | Foundation Model | AWS | 41.4% | +38.9% | 39.3% | 0.26s | Repo ↗ |
| #29 | TFT | Deep Learning | — | 31.7% | +32.2% | 29.2% | 829s | Repo ↗ |
| #30 | Stat. Ensemble | Baseline | Community | 27.5% | +20.7% | 30.6% | 147s | Repo ↗ |
| #31 | Sundial-Base | Foundation Model | Tsinghua University | 27.4% | +33.4% | 33.6% | 8.01s | Repo ↗ |
| #32 | PatchTST | Deep Learning | — | 26.8% | +30.1% | 25.1% | 664s | Repo ↗ |
| #33 | DeepAR | Deep Learning | — | 26.1% | +29.1% | 23.6% | 1113s | Repo ↗ |
| #34 | AutoARIMA | Statistical Baseline | — | 23.3% | +21.3% | 21.5% | 20.14s | Repo ↗ |
| #35 | AutoETS | Statistical Baseline | — | 20.3% | -26.8% | 22.5% | 3.48s | Repo ↗ |
| #36 | CatBoost | Tree Ensemble | — | 19.1% | +23.0% | 35.1% | 29.48s | Repo ↗ |
| #37 | LightGBM | Tree Ensemble | — | 17.8% | +21.0% | 33.1% | 3.03s | Repo ↗ |
| #38 | AutoTheta | Statistical Baseline | — | 15.6% | +5.5% | 19.8% | 3.28s | Repo ↗ |
| #39 | Seasonal Naive | Baseline | Community | 10.3% | 0.0% | 11.8% | 0.48s | Repo ↗ |
| #40 | Naive | Statistical Baseline | — | 6.7% | -45.4% | 10.9% | 0.47s | Repo ↗ |
| #41 | Drift | Statistical Baseline | — | 6.1% | -45.8% | 10.0% | 0.45s | Repo ↗ |
Variable Schema & Summary Distributions
Observed numerical ranges, category cardinalities, missing rates, and sample values across 7,710 rows.
| Col | Variable Name | Type | Missing | Distinct | Summary Stats / Distribution | Sample Values |
|---|---|---|---|---|---|---|
| C | ordersTARGET | NUM | 370 (0.04799%) | 4,426 | [790, 18139] μ=5535.4 σ=2182.7 | 644753385621 |
| A | id | CAT | 0 | 7 | — | Brno_1Brno_1Brno_1 |
| B | timestamp | CAT | 0 | 1,197 | — | 2020-12-05T00:00:002020-12-06T00:00:002020-12-07T00:00:00 |
| D | holiday | NUM | 370 (0.04799%) | 2 | [0, 1] μ=0.0 σ=0.2 | 000 |
| E | school_holidays | NUM | 370 (0.04799%) | 2 | [0, 1] μ=0.0 σ=0.1 | 000 |
| F | shops_closed | NUM | 370 (0.04799%) | 2 | [0, 1] μ=0.0 σ=0.1 | 000 |
| G | winter_school_holidays | NUM | 370 (0.04799%) | 2 | [0, 1] μ=0.0 σ=0.2 | 000 |
| H | blackout | NUM | 370 (0.04799%) | 2 | [0, 1] μ=0.0 σ=0.0 | 000 |
| I | frankfurt_shutdown | NUM | 370 (0.04799%) | 2 | [0, 1] μ=0.0 σ=0.0 | 000 |
| J | mini_shutdown | NUM | 370 (0.04799%) | 2 | [0, 1] μ=0.0 σ=0.0 | 000 |
| K | mov_change | NUM | 370 (0.04799%) | 3 | [0, 1] μ=0.2 σ=0.4 | 000 |
| L | precipitation | NUM | 640 (0.08301%) | 772 | [0, 80] μ=1.6 σ=4.5 | 000.5 |
| M | shutdown | NUM | 370 (0.04799%) | 2 | [0, 1] μ=0.0 σ=0.0 | 000 |
| N | snow | NUM | 640 (0.08301%) | 158 | [0, 19.84000015258789] μ=0.3 σ=1.5 | 000 |
| O | user_activity_1 | NUM | 370 (0.04799%) | 1,135 | [482, 3437] μ=1633.5 σ=784.2 | 233222942305 |
| P | user_activity_2 | NUM | 370 (0.04799%) | 6,147 | [1797, 45769] μ=23477.0 σ=8934.8 | 273922730527255 |
| Q | holiday_name | CAT | 7,492 (0.97173%) | 24 | — | Christmas Eve2nd Christmas DayNew Years Day |
Frequently Asked Questions: Rohlik E-Grocery Daily Fulfillment Orders
Common questions regarding the Rohlik E-Grocery Daily Fulfillment Orders dataset, machine learning task formulations, and in-browser tabular inference.
What is the forecasting horizon for Rohlik E-Grocery Daily Fulfillment Orders?
The dataset evaluates a 61-day forward forecast horizon at a daily frequency, matching the real-world operational planning window required by e-grocery warehouse logistics for seasonal labor scheduling and supplier agreements.
Can I run zero-server forecasts on this dataset in Google Sheets?
Yes, Carla enables zero-server in-context time series forecasting directly in spreadsheets using Mistral's Timestral foundation model, processing historical order lags and calendar covariates entirely in-browser.
Which models perform best on Rohlik E-Grocery Daily Fulfillment Orders?
Foundation models such as Mistral's Timestral, MOIRAI, and gradient boosted tree ensembles (LightGBM/CatBoost) equipped with holiday calendar features outperform classical ARIMA and exponential smoothing baselines by effectively capturing non-linear calendar closures and weather shocks.
Test TabICLv2 on Rohlik E-Grocery Daily Fulfillment Orders Yourself
Open the pre-loaded Google Sheet and let Carla configure the target and task for local, zero-cloud tabular machine learning.
Launch Carla
Click once to open the spreadsheet and Carla side panel together.
Review the Setup
Carla selects the dataset target and task from this page automatically.
Evaluate & Predict
Run predictions and compute metrics with zero server uploads.
Provenance & Attribution
Rohlik Group. (2024). Rohlik Orders Forecasting Challenge. Kaggle.