Hospitality intelligenceHospitality intelligence / forecasting
MISE
A reproducible hospitality-proof with synthetic data that translates reservations, POS, tables, weather context and staff into a controllable daily plan — with the manager as the decision maker.
Interactive proof of conceptThe case in 30 seconds
MISE
- Problem
- A reproducible hospitality-proof with synthetic data that translates reservations, POS, tables, weather context and staff into a controllable daily plan — with the manager as the decision maker.
- Solution
- A reproducible hospitality-proof with 546 synthetic days, 65 unseen test days, table model, staff scheduling and fail-closed data protection.
- Role and focus
- Data model, forecast core, operation UX, integrity
- Status
- Interactive proof of concept
- Result
- 12-route demo, backtest, table and staff plan
Live lab / try it yourself
Enjoy a catering day
Choose a scenario and see how reservations, weather, tables, staff and turnover combine to create an actionable daily plan. This case interaction is a scenario simulator; the fullMISEdemo uses the validated data layer.
Brand vision / preparation as a decision system
Everything is ready before the service starts
MISErefers directly toMISE-en-place: the catering language for good preparation. The logo connects table, timing and central decision without the maritime association of the previous work name.
Operating model / floor and capacity
The floor becomes legible before service
The forecast does not end with a number. The designed table plans make visible how capacity, routing and preparation come together in a verifiable work plan.
Designed demo material from theMISEZwolle setup. The plans show the operational structure; the forecast uses reproducible synthetic data.
Forecasting on the floor
From forecast to work plan
The outcome is not “82% pressure”.MISEsays when the crowds will come, where guests are likely to want to sit, which preparations require extra time and which signals are uncertain.
Reservations, cash register and planning
Reproducible covers, turnover per hour, no-shows, dishes, employees and preparation form the testable basis.
Weather, roadwork and events
The full demo uses current Open-Meteo context. Permits, closures and events are designed integration routes, not connected feeds.
Competition without false security
Deals, reviews and tourism data remain integration options with source quality and local relevance; they are not presented as live truth.
Play through the day
Four scenarios, four different workplaces
Click on a scenario. The forecast not only changes the expected turnover, but also the concrete preparation and the certainty with whichMISEadvises.
Interactive daily moments
Expected covers per hourBandwidth and deployment time
Scale up the team
Scenario outcome184 coversBandwidth 163 to 205
Substantiated trust83%3 of 4 sources active
Reservations determine the basis; weather and local context drive terrace, peak and preparation.
Prepare the terrace first83% confidence · 3 active sources
Covers184+22% vs normal
Peak13:10terrace 76%
Extra starts2 × 60m09:30 patio crew
09:30
Cushions, umbrellas and outdoor station
11:00
Extra cold drinksMISE-en-place
12:30
Second runner to terrace
16:00
Reforecast with current covers
Inner route and quick changeover88% confidence · 3 active sources
Covers132terrace uncertain
Rain14:05± 25 minutes
OccupancyCompact1 late shift postponeable
10:00
Open completely inside
12:00
Limited use of terrace
13:35
Pillows in signal
15:00
Replenish hot drinks
Early family crowds expected72% confidence · 3 active sources
Covers211peak shifts
Passers-by+1.4k125m radius
MenuQuickshorten the lunch menu
09:00
Extra children's set and high chairs
10:30
Grab-and-go station ready
11:40
First peak before reservations
14:30
Recalibrate stock and waiting time
Car inflow lower, local guests stable68% confidence · 3 active sources
Covers119−18% forecast
Walk-ins−31%car traffic
Team−1 shiftstandby at 4:00 PM
08:30
Publish accessibility message
10:00
Make the cycle route visible
12:00
Do not release additional purchases
16:00
Reassess reservations
Human control Scenario not yet applied to planning, purchasing or team.
Design boundary: the numbers are demonstration dates. In a pilot, source quality, local influence and margin of error are calibrated per restaurant; a manager always holds the final personnel and purchasing decisions. Open the full 18-month data and personnel proof.
Forecasting with evidence
Test before adjusting
MISEnot only makes a forecast visible, but also compares it with a fixed test period. The system then translates that outcome into capacity, routing and preparation on the floor.
323 training days and 65 unseen synthetic testing days, with seven forecast vintages and protection against target leakage.
Compared to 18.65% for the four-week baseline. This validates the reproducible proof, not the commercial accuracy of a restaurant pilot.
Table capacity, routing and staff deployment are included as controllable source objects in the work plan.
SHA-256, release version and stale source check block a damaged or outdated data layer before use.
My contribution: data model, forecast core, table and staff logic, twelve-route product architecture, interface, integrity layer, test chain and visual system development.
Range: the backtest is a controlled proof with demonstration data; it supports a decision, but does not replace a local calibration or manager decision.
Data quality
Not every source deserves the same trust
Each input is given a separate score for topicality, local relevance, historical predictive value and availability.
Simulated operational data
Reservations and checkout form reproducible test data and are calibrated per hour and type of day.
Public local signals
Live weather context is connected; permits and traffic remain investigated integration routes until a source contract is in place.
Market and competition data
Reviews and deals are possible signals, not truth. They are only used if source quality and local coherence are demonstrable.
Prepare the terrace first
Weather and preparation determine the opening.
Lunch peak
Reservations and passers-by control the floor.
Reforecast
Current covers replace old assumptions.
Evening plan
Team and stock follow the latest context.
Independent concept case with demonstration data. Photography: Bayram Yalçın and Seher Doğan via Pexels. The restaurant is fictional; brand and interface are designed for this portfolio case.
The flagship series
Four other forms of intelligent experience
From freight and ownership experience to evidence and spatial magic.
AI / logistics / operations
NEXUS
A customer and operations platform that brings together booking, documents, tracking, deviations and communication.
Digital twin / luxury experience
AURELIA 67
A private digital twin that connects form, materials and ownership experience.
AI product/decision support
Stradviso Decision Studio
Controlled AI decision system that converts customer questions into traceable options, human approval and audit trail.
Thailand / property intelligence
DOMORA
A multilingual real estate toolset for the Chiang Mai market that connects listings, local housing requirements, map search, matching, seller intake and social content.