π Lecture Summary
Learning Objectives
The decision process is fundamental to management in both manufacturing and service organizations. The decision-making process involves six important steps: (1) specify objectives and criteria for decision making, (2) develop alternatives, (3) analyze and compare alternatives, (4) select the best alternative, (5) implement the chosen alternative, and (6) monitor the results to ensure desired results are achieved.
The Operations Manager identifies criteria by which proposed solutions will be judged. Common criteria relate to costs, profits, return on investment, productivity, risk, company image, and impact on demand. The ideal aims are: costs should decrease and profits should increase; return on investment should increase along with productivity; risk should decrease while company image increases; and demand should increase for the product or service.
The Decision Process Example
The CEO of ABC Corporation asks the VP Operations to help the Board of Directors decide whether to introduce a new automobile model. The new model has these effects: costs decrease by 15%, profits increase by 2%, return on investment stays the same, productivity decreases by 5%, risk increases by 5%, company image may increase or decrease, and demand may increase or decrease.
π Risk Averse Manager: A manager who would forego the new project due to increased risk. π Risk Taker: A manager who would proceed with the new project despite increased risk.
π‘ Why this matters: These factors alone do not present the overall big picture. In practical situations, decisions are based on important factors like ROI, productivity, utilization of available resources, profits and costs in line with organizational strategy, and mapping the organization relative to its competitors.
Causes of Poor Decisions
Unforeseeable and uncertain circumstances can lead to mistakes in decision making. The remedial action is to establish a STEERING COMMITTEE comprising senior management to review the whole process and monitor the decision steps.
Decision Environments
There are three degrees in decision environments:
-
Certainty: Means that relevant parameters such as costs, capacity, and demand have known values.
- Example: Profit per unit is Rs. 50, and you have an order for 2000 units. The decision is under certainty because all parameters are known.
-
Risk: Means that certain parameters have probabilistic outcomes.
- Example: There is a 25% chance of demand of 2000 units, 50% chance of demand of 1000 units, and 25% chance of an order of 500 units.
-
Uncertainty: Means that certain parameters have various possible future events with no available data.
- Example: There is no available data of demand forecasts, meaning the parameters necessary for decision making are absent.
DECISION THEORY
Decision Theory is a general approach to decision making in Production Operations Management.
Decision theory consists of three elements:
- A set of possible outcomes exist that will have a bearing on the results of the decision.
- A list of alternatives to choose from.
- A known payoff for each alternative under each possible future condition.
An operations manager needs to employ these five steps:
- Identify a set of possible future conditions called state of nature (including low, high, medium demand patterns and competitor's new product introductions).
- Develop a list of alternatives, one of which may be to do nothing.
- Determine or estimate the payoff associated with each alternative for every possible future condition.
- If possible, estimate the likelihood of each possible future condition.
- Evaluate alternatives according to some decision criterion (e.g., maximize expected profit) and select the best alternative.
PAY OFF TABLE
A payoff table summarizes the information of a decision and captures the expected payoffs under various possible states of nature.
π Example: Setting up a pharmaceutical factory. If we build a small facility, the return remains the same whether demand is low or high. The medium facility indicates constant return on moderate and high demand. If we build a large facility, the return would only be good if we have high demand.
| Alternatives | Possible Future Demands |
|---|---|
| Low | |
| Small Facility | Rs. 10 M |
| Medium | Rs. 5 M |
| Large | Rs. 1 M |
The states of nature are very important for decision making.
Decision Making under Certainty
Decision making under certainty is always simple but never available to managers in practice.
π Process: It is known with certainty that demand will be low, moderate, or high. We simply select the best or highest payoff for all states of nature. In the example, if demand is known to be high, we select Large Facility (Rs. 15 M) as it has the highest payoff.
Decision Making under Uncertainty
In the absence of clear information, an operations manager needs to carry out decision making under uncertainty. This is the usual pattern when managers face a dilemma to evaluate alternatives. Four approaches are used:
Maximin
π Maximin: Determines the worst payoff for each alternative; the operations manager chooses the best of the worst alternatives. It is a pessimistic approach that ensures a guaranteed minimum.
π Process: For each alternative, find the minimum (worst) payoff, then select the alternative with the highest of these minimums.
π Example: Using the payoff table:
- Small Facility: worst = Rs. 10 M
- Medium: worst = Rs. 5 M
- Large: worst = Rs. 1 M
- Select Small Facility (Rs. 10 M) as it has the best worst payoff.
Maximax
π Maximax: Determines the best possible outcome and chooses the alternative with the best possible payoff. It does not take into account any other alternative except the best payoff. This is an optimistic approach (go for it strategy).
π Example: Using the payoff table:
- Small Facility: best = Rs. 10 M
- Medium: best = Rs. 12 M
- Large: best = Rs. 15 M
- Select Large Facility (Rs. 15 M) as it has the highest maximum payoff.
Laplace
π Laplace: Determines the average payoff for each alternative and chooses the alternative with the best average. This is a cautious approach that treats the states of nature as equally likely.
π Formula: Average = Sum of all payoffs for an alternative Γ· Number of states of nature
π Example: Using the payoff table:
- Small Facility: (10+10+10)/3 = Rs. 10 M
- Medium: (5+8+12)/3 = Rs. 8.33 M
- Large: (1+2+15)/3 = Rs. 6 M
- Select Small Facility (Rs. 10 M) as it has the highest average.
Minimax Regret
π Minimax Regret: Determines the worst regret for each alternative and chooses the alternative with the best worst regret. This approach seeks to minimize the difference between the payoff realized and the best payoff for each state of nature.
π Two-step process:
- Step I: Construct the Table of Opportunity Losses or Regrets by subtracting each column entry from the highest column value (using absolute values).
- Step II: Select the maximum regret value for each row (alternative), then choose the alternative with the smallest maximum regret.
π Example: Using the payoff table
Step I: Construct Regret Table
| Alternatives | Low | Moderate | High |
|---|---|---|---|
| Small Facility | 10-10=0 | 10-10=0 | 15-10=5 |
| Medium | 10-5=5 | 10-8=2 | 15-12=3 |
| Large | 10-1=9 | 10-2=8 | 15-15=0 |
Note: Regrets are absolute values (positive numbers)
Step II: Maximum regret for each alternative
- Small Facility: max(0, 0, 5) = 5
- Medium: max(5, 2, 3) = 5
- Large: max(9, 8, 0) = 9
- Select Small Facility or Medium (both with regret of 5) as they have the smallest maximum regret.
EXPECTED MONETARY VALUE CRITERION
Decision Making under Risk exists in the area between certainty and uncertainty.
π Expected Monetary Value (EMV): Refers to the best expected value among the alternatives. We use the payoff table with probabilities that must add to 1 and be mutually exclusive and collectively exhaustive.
π Formula: EMV = Ξ£ (Probability of state Γ Payoff for that state)
π Example: Using probabilities Low=0.3, Moderate=0.5, High=0.2:
- EV small = 0.3(10) + 0.5(10) + 0.2(10) = Rs. 10 M
- EV medium = 0.3(5) + 0.5(8) + 0.2(12) = Rs. 7.9 M
- EV large = 0.3(1) + 0.5(2) + 0.2(15) = Rs. 4.3 M
- Select Small Facility (Rs. 10 M) as it has the highest expected value.
Expected Value of Perfect Information
π Expected Value of Perfect Information (EVPI): In certain situations, it is possible to ascertain which state of nature will occur with certainty. For example, if you want to construct a restaurant on a motorway highway, you might get a great ROI.
π Formula: EVPI = Expected payoff under certainty - Expected payoff under risk
Visual Tools for Analyzing Decision Problems
Two visual tools used for analyzing decision problems include Decision Trees and Graphical Sensitivity Analysis.
Decision Trees
A decision tree is a schematic representation of alternatives and their possible consequences presented graphically. The diagram resembles a tree and is extremely suitable for analyzing and evaluating situations involving sequential decisions.
π Example: The Pakistani government decides to operate a gas field. Initially, they can exploit 1 million cubic feet of gas, but later studies indicate potential reserves of an additional 10 million cubic feet. As an operations manager, you may prepare a feasibility report to either expand or make a new facility.
Decision Tree Structure:
- The tree is read from left to right.
- Square nodes represent decisions.
- Circular nodes represent chance events.
- Branches leaving square nodes represent alternatives.
- Branches leaving circular nodes represent chance events (states of nature).
Decision Tree Analysis Steps:
- Analyze the decisions from right to left.
- Determine which alternative would be selected for each possible second decision.
- For a small facility with high demand, select the highest payoff and multiply it with the probable outcome. Put a double slash on alternatives with lower value.
- Repeat steps for both low and high demand patterns for the larger facility.
- Determine the product of chance probabilities.
- Determine the expected value of each initial alternative.
- Select the choice with the larger expected value.
π Decision Tree Example Solution:
Option I: Build Small Facility
- Low Demand = 0.4 Γ Rs. 20 = Rs. 8
- High Demand = 0.6 Γ Rs. 50 = Rs. 30
- Expected Value = Rs. 8 + Rs. 30 = Rs. 38
Option II: Build Large Facility
- Low Demand = 0.4 Γ Rs. 45 = Rs. 18
- High Demand = 0.6 Γ Rs. 60 = Rs. 36
- Expected Value = Rs. 18 + Rs. 36 = Rs. 54
Decision: Select the Larger Facility as it has a larger expected value (Rs. 54) than the small facility (Rs. 38).
Sensitivity Analysis
Sensitivity Analysis involves determining the range of probability for which an alternative has the best expected payoff. It is a graphical solution that makes use of algebra and has prime importance in decision making.
Conclusion
Decision making is a critical responsibility that stays with a manager throughout his active professional life. At the start of service, decision making involves low financial impact, but with time, it becomes more critical and highly finance focused. This aspect gives decision making a competitive edge over other tools available to an operations manager. The related field of game theory is often used in conjunction with decision theory.
β Key Takeaways
The six-step decision process (specify objectives, develop alternatives, analyze, select, implement, monitor) forms the foundation of operations management. Decision environments are classified into certainty (known values), risk (probabilistic outcomes), and uncertainty (no available data), each requiring different analytical approaches. Under uncertainty, managers use maximin (pessimistic), maximax (optimistic), Laplace (average), and minimax regret (minimizing opportunity loss) criteria. Under risk, the Expected Monetary Value criterion combines payoff values with probability distributions to select the best alternative. Decision trees provide visual representation for sequential decisions, and the expected value of perfect information helps managers determine the value of obtaining additional information before making decisions.
π§ Quick Revision Questions
- What are the six steps in the decision-making process, and why is the monitoring step important?
- How do the three decision environments (certainty, risk, uncertainty) differ, and what decision criteria are used for each?
- Using the payoff table with Small (10,10,10), Medium (5,8,12), and Large (1,2,15) in millions, calculate the maximin, maximax, and Laplace decisions.
- What is the Expected Monetary Value criterion, and how is it calculated using probabilities of 0.3, 0.5, and 0.2 for the three payoff table alternatives?
- How do you construct and analyze a decision tree, and what is the difference between square nodes and circular nodes?
π Lecture 07 β FORECASTING
π Overview: This lecture introduces the concept of forecasting in operations management, explaining its critical role in business planning and decision-making. It covers the components of demand, applications across different business functions, and introduces web-based forecasting tools like CPFR. Understanding forecasting is essential for managers to anticipate future demand and plan accordingly.
ποΈ Topics Covered
The lecture begins by comparing business forecasting to weather forecasting, explaining that both are educated guesses but essential for planning. It distinguishes between planning the system (long-term) and planning the use of the system (short/intermediate-term). It then covers the definition of a forecast, applications across business functions, demand management (independent vs. dependent demand), the components of demand, and introduces Collaborative Planning, Forecasting, and Replenishment (CPFR) as a web-based forecasting tool.
π Lecture Summary
Introduction
Forecasting demand is like forecasting weatherβsometimes the forecast fails completely, and sometimes it is near the predicted value but still not exact. Scientists often call forecasting an educated guess, but it helps us plan trips, journeys, and for farmers, to plant, harvest, and take precautionary measures. In business, forecasting forms the basis for budgeting and planning for capacity, sales, production, inventory, manpower, purchasing, and more. Forecasting allows the manager to anticipate the future so they can plan accordingly.
Two Major Uses for Forecasts
There are two major uses for forecasts: one is to help the Operations Manager plan the system, and the other is to help them plan the use of the system. These are distinct but closely linked concepts.
- Planning the system refers to long-term plans about the type of products or services to offer, what facilities and equipment to have, where to locate, and so on. This is more of a senior manager's job with an organizational strategy perspective.
- Planning the use of the system relates to short-range and intermediate-range planning, meaning planning inventory, workforce resources, purchasing, production activities, budgeting, and scheduling. This is an operational strategy.
Business Forecasting Scope
Business Forecasting is more than just predicting demand. It is also used to predict profits, revenues, costs, productivity changes, prices and availability of energy and raw materials, interest rates, movements of key economic indicators (GNP, inflation, government loans), and prices of stocks and bonds. Forecasting is not an exact science. Even with computers and algorithms, it cannot make exact predictionsβit requires experience, managerial judgment, and technical expertise. General responsibility lies with the marketing workforce, but no marketing forecast has been created without valuable contribution from the operations side.
π Definition β Forecast: A statement about the future value of a variable of interest such as resource requirements, capacity planning, supply chain management (SCM), and product or service demand.
Forecasts affect decisions and activities throughout an organization:
- Accounting, finance
- Human resources
- Marketing
- MIS
- Operations
- Product/service design
Applications of Forecasts
| Business Function | Application |
|---|---|
| Accounting | Cost/profit estimates |
| Finance | Cash flow and funding |
| Human Resources | Hiring/recruiting/training |
| Marketing | Pricing, promotion, strategy |
| MIS | IT/IS systems, services |
| Operations | Schedules, MRP, workloads |
| Product/service design | New products and services |
Demand Management
Independent Demand: What a firm can do to manage it?
- Either be active or passive
- Can take an active role to influence demand
- Can take a passive role and simply respond to demand
The lecture distinguishes between independent demand (finished goods/services) and dependent demand (raw materials, component parts, sub-assemblies, etc.). These are illustrated with a Bill of Materials (BOM) structure showing how demand for components depends on demand for the final product.
Components of Demand
The components of demand are:
- Average demand for a period of time
- Trend
- Seasonal element
- Cyclical elements
- Random variation
- Autocorrelation
Web-Based Forecasting: CPFR Defined
Collaborative Planning, Forecasting, and Replenishment (CPFR) is a Web-based tool used to coordinate demand forecasting, production and purchase planning, and inventory replenishment between supply chain trading partners. It is used to integrate the multi-tier or n-Tier supply chain, including manufacturers, distributors, and retailers.
CPFR's objective is to exchange selected internal information to provide for a reliable, longer-term future view of demand in the supply chain. CPFR uses a cyclic and iterative approach to derive consensus forecasts.
Steps in CPFR:
- Creation of a front-end partnership agreement
- Joint business planning
- Development of demand forecasts
- Sharing forecasts
- Inventory replenishment
Key assumptions and characteristics of forecasts:
- Assumes a causal system (that the same system that existed in the past will exist in the future), whereas in reality unplanned events happen like tax rate increases, introduction of a competitor's product or service, or natural disasters.
- Forecasts are rarely perfect because of randomness (having no specific pattern). Allowances should be made for inaccuracies.
- Forecasts are more accurate for groups vs. individuals because forecasting errors in a group tend to cancel out forecasting errors for individuals.
- Forecast accuracy decreases as the time horizon increases, indicating it is safer to make short-range forecasts instead of long-term forecasts.
π‘ Why this matters: Understanding these limitations helps managers know when to trust forecasts and how long into the future they can reasonably plan.
β Key Takeaways
Forecasting is an educated guess about future demand that forms the basis for all business planning, from long-term system design to short-term operational scheduling. The two major uses of forecasts are planning the system (long-term, strategic) and planning the use of the system (short/intermediate-term, operational). Forecasts apply across all business functions including accounting, finance, HR, marketing, MIS, operations, and product design. Demand has six components: average, trend, seasonal, cyclical, random variation, and autocorrelationβand can be managed actively (influencing demand) or passively (responding to demand). Finally, forecasts are rarely perfect due to randomness, are more accurate for groups than individuals, and their accuracy decreases as the time horizon increases.
π§ Quick Revision Questions
- What are the two major uses for forecasts in operations management, and how do they differ?
- List the six components of demand.
- What is the difference between independent demand and dependent demand?
- What does CPFR stand for, and what is its primary objective?
- Why are forecasts more accurate for groups of items than for individual items?
π Lecture 8 β Forecasting (Contd.)
π Overview: This lecture continues the discussion on forecasting by outlining the requirements of a good forecast and the step-by-step forecasting process. It then introduces the fundamental types of forecastsβqualitative and quantitativeβand provides a finer classification into judgmental, time series, and associative models, with special emphasis on judgmental methods like the Delphi method and the simple NaΓ―ve forecast.
ποΈ Topics Covered
The lecture begins with the requirements of a good forecast and the forecasting process. It then introduces fundamental types of forecasts (qualitative and quantitative) and a finer classification (judgmental, time series, associative). Characteristics of judgmental forecasts are discussed, including executive opinions, sales force opinions, consumer surveys, outside opinions, and the Delphi method. The lecture concludes with time series analysis and a detailed look at naΓ―ve forecasts, including their advantages and drawbacks.
π Lecture Summary
Requirements of a Good Forecast
A forecast must be timely, meaning the forecasting horizon should provide enough time to implement changes, as capacity cannot be expanded instantly. It should be reliable, working consistently; otherwise, end users will question its purpose. Forecasts must be accurate with a stated degree of accuracy so users understand limitations and can plan for errors. They should be meaningful, expressed in units relevant to the user (e.g., Rupees for financial planners, machine types for project schedulers). Forecasts need to be written/documented to allow measurement of variance between estimate and actual result. Finally, they must be simple to understand and use, not dependent on sophisticated computer techniques or highly qualified technical personnel, as failure here leads to incorrect decisions and less acceptance.
π Definition β Timely Forecast: A forecast that provides enough lead time to implement necessary changes. π‘ Why this matters: Without timeliness, even an accurate forecast is useless because there is no time to act on it.
Steps in the Forecasting Process
The process begins by determining the purpose of the forecastβwhat it is for and when it is requiredβwhich defines the level of detail and resources needed (man, machine, time, capital). Next, establish a time horizon, noting that accuracy decreases as time increases. Then select a forecasting technique (qualitative or quantitative). After that, gather and analyze the appropriate data; the closer the data is to real life, the more realistic the forecast will be. Prepare the forecast, and finally, monitor the forecast to determine if it is fulfilling its purpose, which helps in re-examining the method, assumptions, and data validity and preparing a revised forecast.
Fundamental Types of Forecasts
Qualitative techniques use subjective inputs and no numerical data, relying solely on soft information like human factors, personal opinion, and hunches. These forecasts are often biased and tilted toward what management wants to predict. Quantitative forecasts involve extending historical data or using explanatory variables to predict future demands. They are favored when quality attributes cannot be quantified. In reality, both types need to be used together to develop a judicious and realistic forecast.
Finer Classification of Forecasts
Judgmental forecasts use subjective inputs obtained from sources like consumer surveys, sales staff, managers, and panels of experts; these insights are not publicly available. Time series forecasts use historical data assuming the future will be like the past, developing relationships between variables to predict future values; some smooth out random variations, while others identify and extrapolate specific patterns. Associative models use explanatory variables to predict the futureβfor example, demand for a small car may depend on the increase in petrol or CNG pricesβemploying a mathematical model that relates the predicted variable to predictor variables.
π Definition β Associative Model: A forecasting model that uses one or more explanatory variables to predict the future value of another variable.
Judgmental Forecasts Characteristics
Judgmental forecasts rely solely on judgment and opinion to make forecasts. They are easy to use when there is not enough time for quantitative analysis. They are preferred when the external environment (economic and political conditions) is changing rapidly. They are also used when introducing new products, services, features, or packaging, where historical data does not exist.
Judgmental Forecasts
Executive opinions consist of a group of senior-level managers from different interfaces, used for long-range planning and new product development. The advantage is a collective pool of information from all divisions; the disadvantage is that one person may dominate others, leading to erroneous forecasts. Sales force opinions have the advantage of direct contact with customers, allowing detection of customer plan changes, but suffer from an inability to differentiate between what a customer can do and will do; current sales data can lead to over-pessimistic or over-optimistic forecasts. Consumer surveys are based on samples from potential customers and require skill to develop, administer, and interpret; they often fall victim to consumers' irrational buying behavior. Outside opinion mixes consumer and potential customer views, available via internet, telephone surveys, and newspapers, but its biggest limitation is a fixed format that often fails to quantify exact demand forecasts.
π Definition β Delphi Method: A forecasting technique where managers and staff complete a series of questionnaires, each developed from the previous one, to achieve a consensus forecast. It is commonly used for technological forecasting and is a long-term, one-time activity with the same issues as expert opinion forecasts.
Time Series Analysis
Time series forecasting models try to predict the future based on past data. Managers can pick models based on: time horizon to forecast, data availability, accuracy required, size of forecasting budget, and availability of qualified personnel.
NaΓ―ve Forecasts
NaΓ―ve forecasts are simple to use, have virtually no cost, are quick and easy to prepare, require no data analysis, and are easily understandable. However, they cannot provide high accuracy and can only serve as a standard for accuracy against which other forecasts can be compared.
π Definition β NaΓ―ve Forecast: The simplest forecasting method that assumes the next period's value will equal the most recent period's actual value.
β Key Takeaways
A good forecast must be timely, reliable, accurate, meaningful, documented, and simple to understand. The forecasting process follows six steps: determine purpose, establish time horizon, select technique, gather data, prepare forecast, and monitor. Forecasts are classified into three main types: judgmental (based on subjective input), time series (based on historical data patterns), and associative (using explanatory variables). The Delphi method is a structured judgmental approach using sequential questionnaires for consensus, while naΓ―ve forecasts are the simplest benchmark with virtually no cost but low accuracy. Quantitative and qualitative methods should be used together for realistic forecasts.
π§ Quick Revision Questions
- What are the six requirements of a good forecast mentioned in the lecture?
- List the six steps in the forecasting process in the correct order.
- What is the main difference between qualitative and quantitative forecasting techniques?
- What are the three finer classifications of forecasts, and how do they differ?
- What is the Delphi method, and for what type of forecasting is it commonly used?
π Lecture 9 β Forecasting (Contd.)
π Overview: This lecture continues the study of forecasting by focusing on time series forecasts and the various patterns that can be identified in data over time. It introduces the concept of averaging techniques, including moving averages, weighted moving averages, and exponential smoothing, which are essential tools for predicting future demand based on historical data.
ποΈ Topics Covered
The lecture covers time series forecast components including trend, seasonality, cycle, irregular variations, and random variations. It then introduces forecast variations and techniques for averaging, specifically moving average and weighted moving average, with detailed formulas and solved problems demonstrating how to compute 3-week and 6-week moving average forecasts using actual demand data.
π Lecture Summary
Time Series Forecasts
A time series is a sequence of data points collected at regular intervals over time. The lecture identifies five key behaviors that can be found in time series data: trend, which is a long-term upward or downward movement often related to population shifts, changing incomes, and cultural changes; seasonality, which are short-term fairly regular variations related to factors like weather, festive holidays, and vacations, commonly seen in supermarkets, restaurants, theatres, and theme parks; cycle, which are wavelike variations of more than one yearβs duration caused by political, economic, and even agricultural conditions; irregular variations, caused by unusual circumstances such as severe weather, earthquakes, worker strikes, or major changes in product or service β these do not capture or reflect the true behavior of a variable and can distort the overall picture, so they should be identified and removed from the data; and random variations, which are caused by chance and are in reality the residual variations that remain after the other behaviors have been identified and accounted for.
π‘ Why this matters: Identifying these components correctly allows a forecaster to choose the appropriate forecasting method and avoid being misled by noise or unusual events in the data.
Forecast Variations
The lecture introduces techniques for averaging as methods to smooth out variations in data and produce more reliable forecasts. The three main techniques covered are: Moving average, Weighted moving average, and Exponential smoothing.
Moving Average
A moving average is a technique that averages a number of recent actual values, updated as new values become available. The simple moving average model assumes an average is a good estimator of future behavior.
π Definition β Moving Average: A forecasting method that computes the average of a specified number of the most recent actual data points and uses that average as the forecast for the next period.
π Formula: Fβ = (Aβββ + Aβββ + Aβββ + ... + Aβββ) / n β The forecast for the coming period (Fβ) is the average of the actual demands (A) from the past "n" periods, where n is the number of periods to be averaged.
Weighted Moving Average
A weighted moving average gives more weight to more recent values in a series when computing the forecast. This allows the forecast to respond more quickly to recent changes in demand.
Simple Moving Average Problem (1)
π Example: What are the 3-week and 6-week moving average forecasts for demand, assuming you only have 3 weeks and 6 weeks of actual demand data for the respective forecasts?
Given data for weeks 1-6: Week 1=650, Week 2=678, Week 3=720, Week 4=785, Week 5=859, Week 6=920
Solution:
-
3-week moving average: Fβ = (650+678+720)/3 = 682.67 (forecast for week 4)
-
Fβ = (678+720+785)/3 = 727.67
-
Fβ = (720+785+859)/3 = 788.00
-
Fβ = (785+859+920)/3 = 854.67
-
6-week moving average: Fβ = (650+678+720+785+859+920)/6 = 768.67 (forecast for week 7)
The complete table shows:
| Week | Demand | 3-Week | 6-Week |
|---|---|---|---|
| 1 | 650 | ||
| 2 | 678 | ||
| 3 | 720 | ||
| 4 | 785 | 682.67 | |
| 5 | 859 | 727.67 | |
| 6 | 920 | 788.00 | |
| 7 | 850 | 854.67 | 768.67 |
| 8 | 758 | 876.33 | 802.00 |
| 9 | 892 | 842.67 | 815.33 |
| 10 | 920 | 833.33 | 844.00 |
| 11 | 789 | 856.67 | 866.50 |
| 12 | 844 | 867.00 | 854.83 |
Simple Moving Average Problem (2)
π Example: What is the 3-week moving average forecast for this data? Assume you only have 3 weeks and 5 weeks of actual demand data for the respective forecasts.
Given data: Week 1=820, Week 2=775, Week 3=680, Week 4=655, Week 5=620, Week 6=600, Week 7=575
Solution:
-
3-week moving average: Fβ = (820+775+680)/3 = 758.33
-
Fβ = (775+680+655)/3 = 703.33
-
Fβ = (680+655+620)/3 = 651.67
-
Fβ = (655+620+600)/3 = 625.00
-
5-week moving average: Fβ = (820+775+680+655+620)/5 = 710.00
-
Fβ = (775+680+655+620+600)/5 = 666.00
β Key Takeaways
Students must remember that time series data contains five distinct components β trend, seasonality, cycle, irregular variations, and random variations β and that irregular variations should be identified and removed from data before forecasting. The moving average technique is a simple but powerful forecasting method that averages a specified number of recent actual values, with the formula Fβ = sum of past n actuals divided by n. The number of periods (n) selected significantly affects the forecast: a smaller n makes the forecast more responsive to recent changes, while a larger n makes it smoother but slower to react. Weighted moving averages improve upon simple moving averages by assigning different weights to different periods, typically giving more weight to recent data. For exam calculations, students must be able to compute moving average forecasts manually and understand that the forecast for period t uses actual data from periods t-1 back to t-n.
π§ Quick Revision Questions
- What are the five components of a time series forecast, and which one should be removed from data before forecasting?
- What is the formula for a simple moving average, and what does each variable represent?
- For the demand data 820, 775, 680, 655, 620, what is the 3-week moving average forecast for week 6?
- How does increasing the number of periods (n) in a moving average affect the forecast's responsiveness to recent changes?
- What is the key difference between a simple moving average and a weighted moving average?
π Lecture 10 β Forecasting (Contd.)
π Overview: This lecture continues the study of forecasting methods, covering advanced techniques including weighted moving averages, exponential smoothing, trend analysis, and associative forecasting. It emphasizes practical application through numerical examples and introduces key measures for evaluating forecast accuracy.
ποΈ Topics Covered
The lecture covers weighted moving average with multiple examples, exponential smoothing models with different alpha values (0.10, 0.60, 0.5, 0.1, 0.4), common nonlinear trends including parabolic and exponential growth, linear trend equations and calculations, associative forecasting using regression, and forecast accuracy measures including MAD, MSE, and MAPE.
π Lecture Summary
Weighted Moving Average Model
The weighted moving average assigns different weights to past data points, with more recent periods typically receiving higher weights. The weights must sum to one.
π Definition β Weighted Moving Average: A forecasting method where each historical data point is assigned a weight, and the forecast is the sum of weighted values. Formula: Ft = w1A1 + w2A2 + ... + wnAn, where βwi = 1
π Formula: Ft = w1A1 + w2A2 + ... + wnAn β Forecast equals the sum of each period's actual value multiplied by its assigned weight
π Example: Given weekly demand (Week 1: 650, Week 2: 678, Week 3: 720) with weights (t-1: 0.5, t-2: 0.3, t-3: 0.2), forecast for Week 4 = 0.5(720) + 0.3(678) + 0.2(650) = 693.4
π Example: Given weekly demand (Week 1: 820, Week 2: 775, Week 3: 680, Week 4: 655) with weights (t-1: 0.7, t-2: 0.2, t-3: 0.1), forecast for Week 5 = 0.1(755)+0.2(680)+0.7(655) = 672
π‘ Why this matters: More weight is given to most recent values, making the forecast more responsive to recent changes.
Exponential Smoothing Model
Exponential smoothing is a sophisticated weighted averaging method where the forecast is adjusted based on the previous period's forecast error.
π Definition β Exponential Smoothing: A forecasting technique that uses a smoothing constant (Ξ±) to weight recent observations more heavily, with the formula Ft = Ft-1 + Ξ±(At-1 - Ft-1)
π Formula: Ft = Ft-1 + Ξ±(At-1 - Ft-1) β New forecast equals old forecast plus alpha times the previous forecast error
Where: Ft = Forecast value for coming time period, Ft-1 = Forecast value in 1 past time period, At-1 = Actual occurrence in 1 past time period, Ξ± = Alpha smoothing constant
π Example: Given weekly demand data and initial F1 = D1 = 820, with Ξ± = 0.10:
- F2 = 820 + 0.10(820-820) = 820
- F3 = 820 + 0.10(775-820) = 815.50
- F4 = 815.50 + 0.10(680-815.50) = 801.95
- Continuing through Week 10: F10 = 776.69
π Example: Same data with Ξ± = 0.60:
- F2 = 820 + 0.60(820-820) = 820
- F3 = 820 + 0.60(775-820) = 793
- F4 = 793 + 0.60(680-793) = 725.20
- F10 = 756.28
π Example: With Ξ± = 0.5, F1 = D1 = 820: F2 = 820, F3 = 820+0.5(775-820) = 797.75, F4 = 797.75+0.5(680-797.75) = 738.88, F5 = 738.88+0.5(655-738.88) = 696.94
π Example 3: With Ξ± = 0.1 (initial = 42): Period 3 forecast = 41.80, error = 43-41.80 = 1.20; Period 4 = 41.92, error = 40-41.92 = -1.92; through Period 12 = 41.73
π Example 3: With Ξ± = 0.4 (initial = 42): Period 3 forecast = 41.20, error = 43-41.20 = 1.80; Period 4 = 41.92, error = 40-41.92 = -1.92; through Period 12 = 40.92
π‘ Why this matters: Higher alpha values make forecasts more responsive to recent changes but less stable.
Common Nonlinear Trends
Nonlinear trends include parabolic and exponential growth patterns that cannot be accurately modeled with straight lines.
π Definition β Parabolic Trends: Trends that are concave upward or downward, representing a quadratic function where the arms widen as values increase
π Definition β Exponential Growth: A nonlinear trend where values increase at an increasing rate over time
Linear Trend Equation
The linear trend equation uses time as the independent variable to forecast future values along a straight line.
π Formula: Ft = a + bt β Forecast for period t equals the intercept (a) plus the slope (b) times the time period
Where: Ft = Forecast for period t, t = Specified number of time periods, a = Value of Ft at t = 0, b = Slope of the line
π Formula for b: b = [nβ(ty) - βtβy] / [nβtΒ² - (βt)Β²] β Calculates the slope using sums of time and actual values
π Formula for a: a = βy/n - b(βt/n) β Calculates the intercept by subtracting slope times average time from average actual value
π Example: Given Week (1,2,3,4,5) and Sales (150,157,162,166,177): βt = 15, βy = 812, βtΒ² = 55, βty = 2499, n = 5
- b = [5(2499)-15(812)] / [5(55)-225] = (12495-12180)/(275-225) = 315/50 = 6.3
- a = 812/5 - 6.3(15/5) = 162.4 - 18.9 = 143.5
- Result: y = 143.5 + 6.3t
Associative Forecasting
Associative forecasting uses predictor variables to forecast values of interest through regression analysis.
π Definition β Predictor Variables: Variables used to predict values of the variable of interest
π Definition β Regression: A technique for fitting a line to a set of points
π Definition β Least Squares Line: The line that minimizes the sum of squared deviations around the line
Forecast Accuracy
Forecast accuracy is measured using three common error metrics that quantify the difference between actual and predicted values.
π Definition β Error: The difference between actual value and predicted value
π Definition β Mean Absolute Deviation (MAD): Average absolute error
π Definition β Mean Squared Error (MSE): Average of squared errors
π Definition β Mean Absolute Percent Error (MAPE): Average absolute percent error
β Key Takeaways
The weighted moving average gives more importance to recent data through assigned weights that must sum to one. Exponential smoothing adjusts forecasts based on previous period errors using alpha values between 0 and 1, where higher alpha values create more responsive but less stable forecasts. Linear trend equations (yt = a + bt) are calculated using formulas for slope (b) and intercept (a) from time series data. Nonlinear trends include parabolic and exponential growth patterns that require different modeling approaches. Forecast accuracy must be evaluated using MAD, MSE, and MAPE to determine which forecasting method performs best.
π§ Quick Revision Questions
- What is the formula for calculating a weighted moving average forecast, and what constraint must the weights satisfy?
- In exponential smoothing, how does changing the alpha value from 0.10 to 0.60 affect the forecast's responsiveness to actual changes?
- What are the formulas for calculating the slope (b) and intercept (a) in the linear trend equation Ft = a + bt?
- Given the data: Week 1 = 150, Week 2 = 157, Week 3 = 162, Week 4 = 166, Week 5 = 177, what is the linear trend equation?
- What are the three main measures of forecast accuracy discussed in this lecture, and what does each measure?
π Lecture 11 β Product & Service Design
π Overview: This lecture introduces the design aspect of operations, emphasizing that products and services are inherently intertwined and must be designed together. It covers the strategic importance of product/service design, key design activities, reasons for design, objectives, and the critical legal, ethical, and environmental considerations that operations managers must address.
ποΈ Topics Covered
The lecture begins by establishing that products and services complement each other and are found in combination in nearly every organization. It then explains the strategic importance of product/service design, major factors in design strategy, specific design activities, reasons for initiating design, objectives of design, and steps in the design process. Finally, it covers legal, ethical, and environmental issues, along with guidelines designers should adhere to.
π Lecture Summary
Product and Service Design Together
Product and Service Design form the very basis of the design aspect of operations. Products and services complement and supplement each otherβa service organization can provide products, and a manufacturing organization can provide services. For example, a cardiologist performing angioplasty provides a service but also supplies heart valves (a product). Similarly, a bank provides financial services but uses cheques (products), and a university provides education services supplemented by books and CDs.
π‘ Why this matters: Understanding that products and services are found in combination helps operations managers design integrated systems that deliver value to customers.
Importance of Product/Service Design
Product/Service design plays a strategic role in helping an organization achieve its goals. A good design ensures customer satisfaction, quality, and lower production costs. Poor design leads to lack of customer interest, poor sales, and can endanger customers' lives. Operations managers must question the safe operations of products or services to safeguard the organization from product or service liability.
Major Factors in Design Strategy
When designing a product or service, organizations must consider these factors:
- Cost
- Quality
- Time-to-market
- Customer satisfaction
- Competitive advantage
A good product or service can be produced at an economical cost with increased quality and less time to market when the organization aims for customer satisfaction, which often results in competitive advantage and increased revenues.
Product or Service Design Activities
When an organization designs a new product/service or refines an existing one, it must follow these activities:
- Translate customer wants and needs into product and service requirements
- Refine existing products and services
- Develop new products and services
- Formulate quality goals
- Formulate cost targets
- Construct and test prototypes
- Document specifications
For example, an automobile manufacturer wanting to produce a fuel-efficient car must refine its existing product, improve quality, reduce costs, construct a prototype, evaluate its performance for robustness, and document specifications along with test results.
Reasons for Product or Service Design
Organizations consider both external and internal reasons for design:
- Economic
- Social and demographic
- Political, liability, or legal
- Competitive
- Technological
The end result should always be an improved, safe, and reliable product that brings revenue and competitive advantage.
Objectives of Product and Service Design
The primary focus is customer satisfaction, with secondary focus on improved function, increased revenues/profits, quality, and cost reduction. Current trends emphasize attention to visual appearance, ease of production/assembly, and ease of maintenance/service. The design department must consider the organization's capabilities in designing goods and services.
Steps in the Design Process
Most organizations follow these steps (not necessarily in order):
- Motivation: Achievement of goals; for mature organizations, includes government regulations, competitive pressure, customer needs, and new technologies.
- Customers: Valuable inputs are essentialβfailure to satisfy customers leads to losing ground to competitors.
- R&D: Research and Development departments generate new ideas for existing or new products/services. Activities are ITERATIVE and employ feedback from customers and operations.
- Competitors: REVERSE ENGINEERINGβdismantling and inspecting a competitor's product to improve one's own product.
- Forecast Demand: Demand for the company's new product or service.
- Manufacturability: Ease of fabrication or assembly, directly affecting cost, quality, and productivity.
- General Considerations: Design, production/operations, and marketing departments work closely together, sharing customer feedback, quality issues, and operations bottlenecks. Legal/regulatory issues and Product Life Cycle Issues must also be addressed.
Legal, Ethical, and Environmental Issues
Organizations operate within a three-dimensional framework:
1. Legal
Operations Managers must understand governmental regulations (federal, provincial, or district) and industrial/service sector obligations.
- FDA, OSHA, CBR: FDA (Federal Drug Agency), OSHA (Occupational Safety Hygiene Administration), CBR (Center Board of Revenue) monitor organizations.
- Product liability: Manufacturer is liable for injury or damages caused by a faulty product.
- Uniform commercial code: Products carry an implication of merchantability and fitnessβa product must be usable for its intended purpose. Example: A non-uniform electricity cable could cause electric shock.
π Definition β Product Liability: A manufacturer being liable in case of an injury or damages caused by a faulty product.
2. Ethical
Operations Managers are under contractual agreement not to exhibit unethical behavior. Releasing products with defects should be disclosed to customers.
3. Environmental
Operations Managers must work within environmental laws. EPA (Environmental Protection Agency) is active in all countries including Pakistan. A CEO can be jailed for failure to comply. The design side must ensure no design seriously jeopardizes the organization's standing toward the environment.
Designers of Product/Service should adhere to Guidelines
These guidelines form the basis of an organization's design strategy:
- Produce designs consistent with company goals: An economical model replaced with a luxurious model may lose the existing customer base.
- Give customers the value they expect: Reliability, safety, endurance, aesthetic, and quality dimensions are what customers seek.
- Make health and safety a primary concern: Green rickshaws are an example of considering user and operator health and safety.
- Consider potential harm to the environment: New products should be better than existing ones and aid environmental protection. Automobile manufacturers are using hybrid models, and steam-operated cars may be available in 5 years.