Forecasting
For Next Week Read five articles under the Additional Required Readings/Supply Chain Management folder on CULearn Quiz 7 on Wednesday will cover these readings
It's tough to make predictions, especially about the future. Forecasting Basics It's tough to make predictions, especially about the future. Yogi Berra
Forecasts are usually wrong Forecasts should contain error measure Forecasting Basics Forecasts are usually wrong Forecasts should contain error measure Aggregate forecasts are more accurate The longer the horizon, the lower the accuracy
Costs of Forecasting Errors IBM sells out new PC. Shortage may cost millions Dell stock plunges, sharply off in its forecast of demand Liz Claiborne said earnings decline is a consequence of excess inventories Toyota believes it can save $100M with accurate ordering and inventory management
Qualitative - Subjective Forecasting Basics Qualitative - Subjective Expertise Based The “Sage” Sales Force Customer Surveys
Quantitative – Objective Forecasting Basics Quantitative – Objective Causal Models Time Series Models
Principal Tool: Regression Analysis Causal Models Use when historical data are available and when the relationship between the dependent (Y) and independent variable(s) (X) can be identified Causal methods are good at predicting turning points in demand and for longer range forecasting Principal Tool: Regression Analysis
Regression Analysis Example: Forecasted value is calculated as follows: Car Value (Y) = Car Price + (-Annual Depreciation * Age) The difference between the actual value and the forecasted value is the residual
Causal Models The goal of regression analysis is to determine the values of the parameters that minimize the sum of the squared residual values for the set of observations. This is known as a “least squares” regression fit.
Prediction based exclusively on previously observed values Time Series Models Prediction based exclusively on previously observed values General Idea: Detect Patterns! Short Term Demand Prediction Prevalent Tool In Operations Understand the players
Random Trend Seasonality Cyclic Time Series Patterns No Pattern Linear (default) or Nonlinear Seasonality Repetition at Fixed Intervals Cyclic Long Term Economy Understand the players
Random
Increasing Linear Trend
Curve Linear Trend
Seasonal & Increasing Trend
The Basic Concept of Time Series The forecast for period t+1 can be calculated at the end of period t as a simple moving average as follows: Ft+1 = (Sum of last n demands/n) = (Dt + Dt-1 +…+Dt-n+1)/n Understand the players Ft+1: Forecast for the next period t+1, made in the current period t D : Actual Demand n : Number of periods
How to Choose the Right Technique Understand the players Demand predictions are dependent on life cycle
How to Choose the Right Technique Understand the players Product Development Expert Historical comparisons Competitive comparisons
How to Choose the Right Technique Understand the players Product Introduction Market tests Consumer surveys
How to Choose the Right Technique Understand the players Growth Causal models
How to Choose the Right Technique Understand the players Steady State Time series Causal models
How to Choose the Right Technique Understand the players Decline Time series Causal models