Generate series
120 days of trend + weekly seasonal pattern + Gaussian noise with weekday / weekend effects.
Live · Energy · Retail · Operations
Time-series forecasting
Pick one of three synthetic 120-day series. A Holt-Winters exponential smoothing model fits trend and weekly seasonality, then forecasts the next 14 periods with 95% prediction intervals. In-sample MAPE is reported as a sanity check.
Pick a series
Forecast
Pick a series.
How it works
Holt-Winters
Additive exponential smoothing with seasonal periods = 7. statsmodels optimises the smoothing parameters.
Forecast
Project the smoothed trend + seasonal pattern forward by the chosen horizon (1–30 days).
Prediction intervals
Compute the in-sample residual standard deviation, apply ±1.96σ for a 95% confidence band.
In-sample MAPE
Mean absolute percentage error against the training window — a quick sanity metric for how well the model fits the history.
Production swap
For multi-variate series we swap in SARIMAX, Prophet, or DeepAR (with the same response shape). Same chart, same UI, real customer data.
Want this on your real demand / load data?
We build production forecasting systems with your operational data — covariates, calendar effects, hierarchies, and the explainability your planners need to act on the numbers.
Ready to start
Turn one AI use case into measurable production value.
Book a 30-minute consultation. We will walk through the use case, sketch the value case, and tell you honestly whether we can help.