文档介绍:Demand Forecasting
Lectures 2 & 3
Fall 2003
Caplice
1
Agenda
The Problem and Background
Four Fundamental Approaches
Time Series
General Concepts
Evaluating Forecasts – How ‘good’ is it?
Forecasting Methods (Stationary)
Cumulative Mean
Naïve Forecast
Moving Average
Exponential Smoothing
Forecasting Methods (Trends & Seasonality)
OLS Regression
Holt’s Method
Exponential Method for Seasonal Data
Winter’s Model
Other Models
MIT Center for Transportation & Logistics – 2 © Chris Caplice, MIT
2
Demand Forecasting
The problem:
Generate the large number of short-term, SKU
level, locally dis-aggregated demand forecasts
required for production, logistics, and sales to
operate essfully.
Focus on:
Forecasting product demand
Mature products (not new product releases)
Short time horiDon (weeks, months, quarters, year)
Use of models to assist in the forecast
Cases where demand of items is independent
MIT Center for Transportation & Logistics – 3 © Chris Caplice, MIT
3
Demand Forecasting – Punchline(s)
Forecasting is difficult – especially for the future
Forecasts are always wrong
The less aggregated, the lower the accuracy
The longer the time hori zon, the lower the accuracy
The past is usually a pretty good place to start
Everything exhibi ts seasona l ity of some sort
A good forecast is not just a number – i t should
include a range, description of distribution, etc.
Any analytical method should be supplemented by
external information
A forecast for one function in pany might not be
useful to another function (Sales to Mkt to Mfg to Trans)
MIT Center for Transportation & Logistics – 4 © Chris Caplice, MIT
4
Cost of Forecasting vs uracy
Å Overly Naïve Models Æ Å Good Region Æ Å Excessive Causal Models Æ
Total Cost
Cost
Cost of Errors Cost of Forecasting
In F