Croston Model – Forecasting Intermittent Spare Parts Demand

Introduction

Traditional forecasting methods often fail when demand is intermittent, leading to excessive inventory or stockouts. The Croston model addresses this challenge by separating demand size from frequency, providing a more accurate forecast for sporadically used spare parts.

How the Croston Model Works

Croston divides the intermittent demand time series into two components:

1. Demand size (d) – the quantity consumed when an event occurs

2. Inter-demand interval (p) – the time between consecutive demand events

The forecast at time t is calculated as:

Applications

• Ideal for low-volume, critical parts where demand is irregular.

• Reduces forecast bias caused by traditional smoothing methods.

• Supports inventory optimization by accurately predicting average demand rates over time.

Advantages

• Significantly improves forecast accuracy for intermittent demand.

• Easy to integrate into existing inventory planning systems.

• Reduces unnecessary stock while maintaining high service levels.

Implementation Tips

• Use exponential smoothing for updating both demand size and interval.

• Review historical data to identify patterns or trends.

• Combine with safety stock policies to account for variability.