← All methods

Intermittent · Free

Syntetos–Boylan

Croston with a (1 − α/2) bias correction — usual spare-parts default.

How it works

Croston’s size/interval ratio is positively biased. Syntetos–Boylan multiply by (1 − α/2). The rest of the story is Croston: update on hits, not on zeros. Planners often start here for intermittent or lumpy SKUs. Smooth demand should use SES; obsolescence risk should use TSB. Ranking is still walk-forward MASE, not “SBA is always right.”

Sample forecast question

Same spare-part series as Croston. What is the Syntetos–Boylan (bias-corrected) mean weekly demand for the next four weeks?

Step-by-step on these numbers

1. The question

   Same spare-part series as Croston. What is the Syntetos–Boylan (bias-corrected) mean weekly demand for the next four weeks? Sample weeks: W1=0, W2=0, W3=6, W4=0, W5=0, W6=4, W7=0, W8=0, W9=0, W10=8, W11=0, W12=5 (units).

2. Split size and interval

   Nonzero demands: W3=6, W6=4, W10=8, W12=5. Zeros are not averaged in as “small demand” — they only stretch the interval.

3. Updates on hits only

   SES with α = 0.050 (grid on {0.05…0.40}) updates size z and interval p when y_t > 0. After the sample: z = 5.955, p = 2.997.

4. Bias correction

   SBA multiplies Croston’s z/p by (1 − α/2) = 0.975. Mean demand = (z/p)×(1 − α/2) = 1.94. MAPE is a poor KPI here; the add-in ranks with MASE.

5. Forecast

   W13: 1.94; W14: 1.94; W15: 1.94; W16: 1.94.

6. Interval

   Residual σ = 2.576. For h = 1 the 95% band is [-3.11, 6.99] around 1.94. The engine labels this a residual-Gaussian heuristic (σ√h), not a simulation interval.

Graph of this sample

Related methods: CrostonTeunter–Syntetos–BabaiSimple exponential smoothing

Educational only. Not investment, weather, or operational advice. In the add-in, rank is rolling-origin MASE — “best supported among candidates on this series.”

Use it in the Excel Add-in