I.
Introduction to Forecasting
Forecasting is the art and science of predicting future events, particularly the demand for
output from the operations function.
Demand may differ from sales, making forecasting essential for developing operating
plans.
II. Forecasting Framework
Differences Between Forecasting, Prediction, and Planning:
o Forecasting: Estimates a future event using past data combined systematically.
o Prediction: Estimates a future event based on subjective considerations.
o Planning: Determines what should happen by outlining actions to achieve a goal.
Forecasting as an Input to Business Planning:
o Marketing: Product, pricing, promotion, and placement planning.
o Finance: Financial planning.
o Operations: Capacity planning, inventory management, and more.
Demand Management: Coordinates and controls demand to optimize production
efficiency and delivery.
Types of Demand:
o Dependent Demand: Linked to the demand of another product.
o Independent Demand: Not derived from another product.
Importance of Forecasting:
o Helps make advanced decisions due to lead time and supply chain planning.
o Essential for strategic planning and decision-making.
III. Forecasting Horizon
Short-term (0-3 months): Used for inventory management and scheduling.
Medium-term (3 months-2 years): Used for production planning, purchasing, and
distribution.
Long-term (2+ years): Used for capacity planning, facility location, and strategic
planning.
IV. Forecasting and Operation System
Forecasting integrates with the operation system through:
o Demand Forecasting: Informs planning for product/process design, equipment
investment, and capacity planning.
o Scheduling: Affects production scheduling and aggregate production planning.
o Control: Aids in production, inventory, labor, and cost control.
V. Characteristics of Demand Over Time (Time Series Components)
Patterns in Time Series:
o Constant: Minimal variation.
o Trend: Persistent upward or downward movement.
o Seasonal: Regular fluctuations within a year.
o Cyclical: Business cycle-influenced patterns over multiple years.
o Random Variation: Erratic, unpredictable fluctuations.
Noise Level in Data:
o Low Noise: Data closely follows the pattern.
o High Noise: Data fluctuates significantly from the pattern.
VI. Useful Forecasting Models
1. Qualitative & Judgment Methods
Used when data is unreliable, unavailable, or for new product introductions.
Types:
o Grass Root: Forecast builds from front-line operators' data.
o Delphi Technique: Anonymous expert consensus through questionnaires.
o Market Surveys: Data collection through panels and questionnaires.
o Life-cycle Analogy: Uses historical products as references.
o Informed Judgment: Expert opinions based on experience and facts.
2. Quantitative Methods
Naïve Method: Assumes next period’s demand equals the most recent period’s demand.
Time Series Methods:
o Simple Mean: Average of all available data.
o Simple Moving Average: Averages several recent periods’ data.
o Weighted Moving Average: Assigns greater weight to recent data.
o Exponential Smoothing: Considers all past observations, with recent ones
weighted more heavily.
o Regression Analysis: Uses past data trends to predict future demand.
Causal Relationship Models: Compare forecasted data against external causal factors
(e.g., newspaper sales vs. population size).
VII. Forecasting Errors
Forecast Error Calculation:
o Formula: et = At - Ft (Actual Demand - Forecasted Demand)
Error Measurement & Importance:
o Helps monitor outliers and detect poor forecasting methods.
o Assists in setting safety stock and adjusting forecasting parameters.
Common Forecast Error Measures:
o Cumulative Sum of Forecast Error (CFE) & Mean Forecast Error (MFE):
Measure forecast bias.
o Mean Square Error (MSE): Measures forecast variance.
o Mean Absolute Deviation (MAD): Measures average forecast error magnitude.
o Mean Absolute Percentage Error (MAPE): Expresses error as a percentage of
actual demand.
o Tracking Signal (TS): Detects bias in forecasting.
VIII. Selecting a Forecasting Method
Factors to Consider:
o Use and Decision Needs:
Higher accuracy required for pricing decisions.
Different methods suit different time horizons.
o Data Availability & Quality: Determines appropriate model selection.
o Data Pattern:
Flat Series: Use Naïve or Simple Exponential Smoothing.
Trending or Seasonal Data: Use trend-adjusted models.
Highly Variable Data: Consider qualitative methods.
o Budget and Personnel Expertise: Simpler models are cost-effective; complex
models require skilled personnel.
IX. Importance of Forecasting
Reduces uncertainty and enhances future planning.
Helps businesses anticipate and manage change.
Improves communication and integration among planning teams.
Supports inventory and capacity management.
Assists in cost projection and budgeting.
Enhances competitiveness, cost reduction, and responsiveness to customer needs.
By understanding these concepts and techniques, you will be well-prepared for your exam.
Review the formulas for quantitative methods and ensure you can interpret forecast error
measures
Bloom's Level 2 Question (Understanding):
Which of the following statements best describes the key difference between qualitative and
quantitative forecasting methods as outlined in the source?
a) Qualitative methods rely solely on historical data, while quantitative methods incorporate
expert opinions.
b) Qualitative methods are used for short-term forecasting, whereas quantitative methods are
used for long-term forecasting.
c) Qualitative methods are based on estimates and opinions, while quantitative methods are
based on data related to past demand.
d) Qualitative methods are more accurate than quantitative methods because they consider
subjective factors.
Calculation Example:
Given the following actual demand for the past three periods:
Period 1: 20 units
Period 2: 25 units
Period 3: 30 units
Using a simple moving average of three periods, what would be the forecast for Period 4?
a) 25 units
b) 25 units [(20+25+30)/3 = 75/3 = 25]
c) 30 units
d) 75 units
Discussion:
Bloom's Level 2 Question Discussion:
The correct answer is (c). The source explicitly states that qualitative and judgment methods
are "based on estimates and opinions." These methods are used when there is a lack of reliable or
relevant data, such as when a new product is introduced. Examples of qualitative methods
mentioned include:
Grass root
Delphi Technique
Market Surveys
Life-cycles (historical) Analogy
Informed Judgment
On the other hand, time series (quantitative/extrapolative models) are defined as methods
"based on data related to past demand" to predict future demand. Additionally, causal
relationship (quantitative) or explanatory models are also mentioned as quantitative
techniques.
The fundamental distinction lies in the type of information used:
Qualitative methods rely on subjective opinions.
Quantitative methods rely on historical data.
Options (a), (b), and (d) present inaccurate descriptions based on the information provided in the
source.
Calculation Example Discussion:
The correct answer is (b) 25 units.
A simple moving average forecast for the next period is calculated by taking the average of the
actual demand over a specified number of the most recent periods. In this case, we are using a
three-period moving average. The formula for the forecast for Period 4 (Ft+1) is:
Substituting the given values:
Therefore, the forecast for Period 4 is 25 units.
The source provides a similar example illustrating the calculation of a simple moving average.
Options (a), (c), and (d) are incorrect based on the simple moving average calculation method
described in the source.