Topic 3 Flashcards
Demand forecasting definitions
Demand = Sales of product; usage of part/material Forecasting = Estimation of a future event
(Prediction!)
Why and how do you demand forecast
Why forecast demand?
• Planning : staffing, resources, purchasing
HOW?
• Current factors and past experiences
Different periods of times for forecasting
• Long term (5 years): For system design
• Medium to short term (1 year or less): System
Demand forecasting: common features
• Assumes causal system (past -> future) - (analyze patterns and events in past to help determine future projection and strategies)
• Rarely perfect
• Less accurate for longer time horizon
• More accurate for groups (errors with average out) vs. individuals
Elements of a good forecast
• Timely
• Accurate
• Reliable
• Meaning full units/ simple to understand and use
• In writing
- cost effective
Demand forecasting process
- Determine the purpose of the forecast Level of detail required, amount of
resources and level of accuracy - Establish a forecast horizon
- Gather and analyze relevant historical data
- Select a forecasting technique
- Prepare a forecast
- Monitor a forecast
Approaches to forecasting
Judgmental
- Non-quantitative analysis of subjective inputs
- Considers “soft” information
- (human factors, experience, instinct)
Quantitative: analyze hard data
- Time series models -> Extension of historical patterns of numerical data
- Associative models -> Equations with explanatory variables to predict the future
Judgemental Methods
- Designing new products, redesigning existing prod., using sales promotions
- Executive opinions
Pool opinions of high-level executives
Long term strategic or new product development - Sales force opinions
Based on direct customer - Consumer survey
Questionnaires and focus groups - Historical analogies
Use demand fir a similar product - expert opinions
Delphi method: iterative questionnaires circulated until consensus is reached
Technologicalcontact forecasting
Time series model
Time series is a time ordered sequence of observations taken at regular intervals of time.
Possible patterns in a time series
Level: Horizontal pattern
Trend: steady upward or downward movement
Seasonality: regular variations (related to time of year, month, week, or day)
Time series models
Cyclic: wavelike variations lasting more than one year
Irregular variations: caused by unusual circumstances, not
reflective of a typical behavior
Random variations: residual variations after all other behaviors are accounted for (noise)
Naive Methods s
• Simple to use and understand
• Very low cost
• Low accuracy
Exponential Smoothing: sophisticated weighted moving average
- new forecast is based on the actual demand and forecast for the pervious period
Exponential Smoothing: subjectively choose smoothing constant alpha
- alpha ranges from 0 - 1
- larger the smoothing constant (alpha) the more responsive the forecast
- use higher value of alpha when demand is increasing
Techniques for Trend - LINEAR TREND
- look at historical data to discover if a term exists
- Involves the development of an equation that describes the trend (presuming a trend is present in the data)
- trend component: linear and non-linear
What can regression analysis be used for
• Regression analysis can be used to Fit a trend line (i.e. find the equation of the straight trend line) to a series of historical data
• Equation: ŷt = a + bt
Two most important factors of choosing a forecasting technique is
- cost
- accuracy
Other factors
- availability of historical data
- Forecasting horizon
- pattern of data
