A Study of Different Croston-Like Forecasting Methods


  • Anders Segerstedt1 (Lulea University of Technology, Sweden)
  • Erik Leven1 (Lulea University of Technology, Sweden)

Croston presented an idea and method to separate ordinary exponential smoothing in to two parts: the time between demand (= withdrawals) and demand size. The forecasts then update only when there is a demand. Since then, modifications and alternatives to Croston’s idea have been suggested, i.e. techniques to handle intermittent demand. We test from the beginning four different suggestions to treat intermittent demand or slow-moving demand. The tests showed that some complementary modifications were interesting to investigate. We compare the different techniques with Mean Squared Error (MSE), Cumulated Forecast Error (CFE) and with a new bias measure “Periods in Stock” (PIS). Our tests show that Croston’s original is to prefer; some techniques overestimate and others underestimate demand in certain circumstances; one technique is not to prefer at all. Underestimate leads to shortages and lost sales; lost income is worse than temporarily increased inventory holding costs. We suggest an indicator or rule: when not to forecast and not to stock at all and only restock when a demand occurs. Finally, we discuss the challenge to implement these techniques in practical inventory control.

Download full PDF Get metrics Rate article