Demand pattern classifier
Paste a demand history — one number per period, zeros included — and find out whether the item is smooth, erratic, intermittent, or lumpy, and which forecasting method actually fits it. Runs entirely in your browser.
Results
Paste your demand history above to classify it.
How the classification works
Two standard measures (Syntetos–Boylan) place every item in one of four quadrants:
- ADI (average demand interval) = number of periods ÷ number of periods with demand. ADI ≥ 1.32 means demand arrives infrequently.
- CV² = squared coefficient of variation of the non-zero demand sizes. CV² ≥ 0.49 means the sizes vary a lot when demand does occur.
- Smooth — regular timing, stable sizes. Classic methods work: exponential smoothing (SES) or ARIMA.
- Erratic — regular timing, volatile sizes. SES-family methods still apply, but expect wide error bands and size buffers accordingly.
- Intermittent — long gaps, stable sizes. Use Croston’s method, which forecasts interval and size separately.
- Lumpy — long gaps and volatile sizes. The hardest quadrant: use SBA (Syntetos–Boylan Approximation) and hold judgment-checked buffers.
Read more in our guides to demand classification and intermittent demand.
NextDemand runs this classification — and the matching forecasting method — automatically for every item in your catalog.