Forecasting Models
Flux employs a hybrid ensemble model to predict demand. We combine two primary statistical approaches:
- SARIMA: This model is excellent at capturing weekly seasonality (e.g., Fridays are busier than Mondays) and long-term trends.
- Prophet: Developed by Facebook, this additive regression model handles outliers and holiday effects robustly. It allows us to incorporate "special events" like Super Bowl Sunday or Valentine's Day.
from flux.engine import Forecaster
# Initialize Prophet with seasonality
model = Forecaster(
seasonality_mode='multiplicative',
daily_seasonality=True,
weekly_seasonality=True
)
# Fit on historical sales data
model.fit(
df=sales_history,
regressors=['rain_mm', 'holiday_flag']
)
# Generate 7-day prediction
forecast = model.predict(horizon=7)External Regressors
We don't just look at your sales. We ingest local weather data (Precipitation, Temperature) and Holiday calendars.
Our Bayesian inference engine calculates the coefficient of rain on your walk-in traffic. If the forecast probability of rain > 60%, Flux automatically dampens the demand prediction, ensuring you don't over-prep for a washout.
Your Data is Isolated.
Flux uses Row Level Security (RLS) at the database level. This means that every query sent to our database is automatically filtered by your unique `tenant_id`.
It is mathematically impossible for another restaurant to see your sales data, recipes, or staff schedules.
Flux connects directly to your existing tech stack.
We support the following APIs for real-time data ingestion.
Square API
Sales, Items, Modifiers
Toast API
Labor Data, Sales Mix
Clover API
Inventory, Order Management
Ready to see it in action?
Experience the power of automated operations. No credit card required.
Start Free Trial