Operation Management Online quiz

Kevin11

Chapter_5.ppt

Home >Business & Finance homework help >Operation Management Online quiz

Forecasting

Chapter 5

5 | *

Learning Objectives

Explain the need for forecasting of independent demand products.
Describe the basic principles of forecasting.
Discuss the fundamental components of demand and types of forecasting methods.
Demonstrate and apply time-series analysis in forecasting.
Evaluate forecast accuracy and determine the best method.
Discuss the application of qualitative and casual methods for forecasting.
Describe forecasting across the broader supply chain.

5 | *

Forecasting

“Forecasting involves using several different methods of estimating to determine possible future outcomes for the business. Planning for these possible outcomes is the job of operations management. Additionally, operations management involves the managing of the processes required to manufacture and distribute products.” Smallbusiness.chron

5 | *

Special K = Special Challenges

Why is it very challenging to make a profit in the market for processed cereals such as Raisin Bran, Cheerios, Rice Krispies or Wheaties?

There is a great deal of competition.

Sales are growing slowly (1-5% yearly).

5 | *

Forecasting Kellogg’s Sales

While Kellogg’s sales are good, forecasting needs to be done in order to plan the number of production and distribution facilities to operate, as well as the size and location of such facilities.
Forecasting affects the scheduling of delivery trucks, labor and marketing campaigns.

Source: © Image Source/Corbis

5 | *

Figure 5.1: Kellogg’s Sales of Medium-Sugar Cereal and Cereal Bars in the United States

Source: Breakfast Cereal, U.S. August, 2007 Supply Structure, Mintel International Group Limited

5 | *

Why We Need to Forecast

Forecasts are vitally important to organizations.
They are used to plan facilities, production schedules, staffing allocation, capacity planning, and other things.
The goal of a business forecast is not to have a perfect forecast but to have a reasonable forecast that helps us plan.

Source: © Image Source/Corbis

5 | *

Discussion Starter

How accurate are forecasts?

5 | *

Principles of Forecasting

Forecasts are wrong.
Forecasts get worse for farther into the future they go.
Aggregated forecasts for product or service groups tend to be more accurate.

a whole formed by combining several (typically disparate) element

formed or calculated by the combination of many separate units or items; total.

form or group into a class or cluster.

Forecasts are not a substitute for derived values.

5 | *

Components of Demand

Demand for products or services consists of several components:

average demand

trend

seasonal component

cyclical component

Autocorrelation

Definition of aggregate: General: Collective amount, sum, or mass arrived-at by adding or putting together all components, elements, or parts of an assemblage ...

random variation

5 | *

Figure 5.2: Components of Demand

5 | *

Forecasting Methods

Time-series analysis: a technique that utilizes past demand data to predict future demand by examining cyclical, trend and seasonal influences
Casual relationships: a technique that identifies a connection between two factors, one that precedes and causes changes in the second or effect factor
Qualitative forecasting: a method of forecasting that is based on subjective factors, estimates and opinions

5 | *

Time-Series Analysis

Is based on historical data and the assumption that past patterns will continue in the future.
Goal: to identify the underlying patterns of demand and develop a model to predict these patterns in the future
Five basic techniques:

Naïve forecast

Estimating the average

Moving averages

Weighted moving average

Exponential smoothing

5 | *

Naïve Forecast

Naïve forecast: a method of forecasting that uses the demand for the current period as the forecast for the next period
The naïve forecast is very simple and low cost to use.
Works best when demand, trend and seasonal patterns are stable and there is relatively little random variation.
The naïve approach is the simplest of all the possible forecasting methods and works particularly well when there is autocorrelation.

5 | *

Estimating the Average

Every series of demand figures includes at least two of the six components of demand:

an average and

random variation.

5 | *

Figure 5.3: Daily
Customers at FoodCo Grocery

5 | *

Moving Average

Moving average: a technique for estimating the average of a demand series and filtering out the effects of random variation
Developing a moving average involves computing the average of n previous periods of demand and then using this as the estimate for the next period of demand.
The average is updated after every period to include the most recent demand data.

5 | *

Moving Average

Ft+1 =

where Dt = actual demand in period t

n = total number of periods in the average

Ft+1 = forecast for period t + 1

5 | *

Moving Average: An Example

Daily Customers at FoodCo Grocery

The moving-average forecast at the end of day 4

for the number of customers on day 5 is

Thus, at the end of day 4, we forecast
that there will be 217 customers on day 5.

217.25

Day	Customers
1	228
2	228
3	225
4	188

5 | *

The Moving-Average Technique

Technique used for estimating the average of a demand series and filtering out the effects of random variation
Involves computing the average of n previous periods of demand and then using this as the estimate for the next period of demand (or a period farther out in the future)
Average is updated after every period to include the most recent demand data

5 | *

Figure 5.4: Comparison of
Moving-Average Forecast with
Two- and Four-Day Moving Average

5 | *

Weighted Moving Average

Weighted moving average: a technique that allows periods to have different weights, with the total weight equaling 1.0 (one)
The benefit of a weighted moving average is that it allows a greater emphasis on the most recent demand than on past demand.

5 | *

Weighted Moving
Average: An Example

Ft+1 = 0.6Dt + 0.25Dt-1 + 0.15Dt-2

F4 = 0.6D3 + 0.25D2 + 0.15D1 = 0.6 * 228 + 0.25 * 228 + 0.15* 225 = 226.2

Day	Customers	Weight
1	228	.6
2	228	.25
3	225	.15

5 | *

Exponential Smoothing

Exponential smoothing: a technique that calculates forecasts by giving more weight to recent demands than to earlier demands
Used mostly because it is often easier to calculate than a weighted moving average and requires less data.

5 | *

Exponential Smoothing

F1 = F t-1 + α(A t-1 – F t-1)

Where: F 1 = the New Forecast

F t-1 = Previous period’s forecast

α = (Alpha) a Smoothing Constant

A t-1 = Previous period’s actual demand

5 | *

Exponential Smoothing: An Example

In January, the clothing store estimates February demand for boot cut jeans to be 350. Actual February demand was 310. Using a smoothing constant chosen by the clothing store management of α = .20, what is the forecast for March demand using the exponential smoothing model?

March = last period’s forecast + α(last period’s actual demand – last period’s forecast)

March = 350 + .20(310 – 350) = 350 - 8 = 342

5 | *

Including a Trend

Weighted moving averages and exponential smoothing will adjust to a trend if high weights are given to more recent periods.
Trend-adjusted exponential smoothing is a method for including a trend component in an exponentially smoothed forecast.

5 | *

Including Seasonality

Many organizations sell products or services that have a seasonal demand.
A seasonal demand is characterized by regular repetition of increases or decreases in demand as measured in time periods of less than a year (quarters, months, weeks, days, or hours).

5 | *

Figure 5.7: An Illustration
of Demand for Mousetraps

5 | *

Seasonal Components

Airline travel in the United States has substantially higher demand in the summer and holidays.
Forecasting at an aggregate level is often easier than forecasting at a more detailed level.

5 | *

Figure 5.8: Airline
Traffic for All U.S. Airlines, 2003–2007

Source: Data drawn from http://www.bts.gov/.

5 | *

Multiplicative Seasonal Method

Multiplicative seasonal method: a forecasting method that calculates seasonal factors that are multiplied by an estimate of the average demand to develop a seasonal forecast
Seasonality is measured as a percentage or seasonal index of the average demand for a particular season, which is used to multiply the average value of the series.

5 | *

Seasonal Indices

Utilized to decompose the data and to include seasonality in the forecast.
Decomposing the data involves separating the data into a seasonal, an average, and a trend component. This is done by dividing each data point by its corresponding seasonal index (i.e., divide July demand by the July index, August demand by the August index, and so on).
The seasonal indices are then combined with data on average demand and the trend component using a two-step process:

Develop trend estimates for the desired periods.

Add seasonality to the trend estimates by multiplying these trend estimates by the corresponding seasonal index.

5 | *

Reducing Seasonality

The highly variable nature of demand for products with high seasonality often influences organizations to be more proactive in trying to reduce seasonality.
One method: advertising during slower-demand periods.
Another method: discounting the product or service in periods of slower demand to increase sales.

5 | *

Figure 5.10: Illustration
of Reducing Seasonality

5 | *

Measuring Errors and
Selecting a Time-Series Method

The best forecast is less wrong than the next best forecast.
All forecasts are wrong to some extent, so how do we choose the “best” forecast?
Generally, managers will examine a range of forecast types over a period of time and choose the one with the least amount of error.

5 | *

Forecast Error

Forecast error: the difference between the forecast and the actual demand for a given period
Forecast errors can be classified as either bias errors or random errors.

Bias errors occur when the forecast is consistently over or under the actual demand.

Random errors result from unpredictable factors and do not exhibit a distinct pattern.

Forecasters try to eliminate as much of both types of error as possible, but some error always remains.

5 | *

Measures of Forecast Error

Calculating forecast = the difference between actual demand and the forecast:

Et = Dt – Ft

Et = Forecast error for period T

Dt = Actual Demand for period T

Ft = Forecast for period T

5 | *

Cumulative Sum of Forecast Errors

The cumulative sum of forecast errors (CFE) measures total forecast error:

CFE = ΣEt

5 | *

Mean Absolute Percentage Error

A measure that reports error in proportion to the demand

5 | *

Figure 5.11: Comparison
of Three Forecasting Methods

5 | *

Additional Forecasting Methods

Qualitative Methods
Casual Methods
Linear Regression

5 | *

Qualitative Methods

Market Research
Delphi method
Sales Force Planning

5 | *

Market Research

Market research: a systematic approach to measuring customer interest in a service or product through data-gathering surveys
Market research is widely used for new products, but it also has a high degree of uncertainty and must be interpreted with caution.

5 | *

Delphi Method

A forecasting method that uses a team of experts to develop a consensus forecast.
The Delphi method is useful for long-range forecasts of demand and technological forecasting.

5 | *

Sales Force Planning

Sales force planning: gathering the opinions of salespeople and managers for a particular product or family of products.
Frequent contact with customers often provides insight into what customers may be considering for the future and also into customer perceptions of the company and its products.

5 | *

Casual Methods

When historical data is available and there is a relationship between the item to be forecasted and some other factor (such as advertising expenditure, sales of another product, or government regulations), then a casual method is used.
Casual methods employ mathematical techniques to relate one or more independent variables to the variable being forecast.

5 | *

Linear Regression

Linear regression: a statistical technique that expresses the forecast variable as a linear function of one or more independent variables

5 | *

Figure 5.12: Linear
Regression Line with Raw Data

5 | *

Linear Regression

Three measures usually reported:

Correlation coefficient: a measure of the strength and direction of the relationship between the independent variable x and the dependent variable y
Coefficient of determination: a measure of the amount of variation in the dependent variable that the regression line explains
Standard error of the estimate: a measure of the distance between the dependent variable and the regression line

5 | *

Supply Chain Forecasting

The biggest gains over the last 10 years have

come with the development of more powerful

information technology.

Can capture and process enormous amounts of data
Can quickly connect numerous players within a supply chain at fairly low cost in terms of time and effort.

5 | *

Information Systems for
Sharing Data Across the Supply Chain

Selecting the “best” technology for a specific business or situation is a continuous challenge.
IT, hardware, and techniques are always changing.
Numerous factors affect the performance of any particular choice.

5 | *

Radio Frequency Identification

Radio frequency identification: a technology that utilizes an integrated circuit and a tag antenna printed on a tag to transmit and record information on a product
RFID addresses two of the key limitations of bar codes:

RFID can allow indirect reads and permit multiple items to be read simultaneously.

RFID has the ability to carry substantially more information than most bar codes, as well as to both read and write information on a tag.

5 | *

Collaborative Planning, Forecasting and Replenishment (CPFR)

CPFR: a group of business processes supported by information technology where supply chain members agree to shared business objectives and measures, develop joint sales and operational plans, and collaborate electronically to generate and revise forecasts and production plans

5 | *

Wal-mart’s Application of RFID and
CPFR: Sharing Data Across the Supply Chain to Maximize Forecast Accuracy

Wal-mart’s IT system has 99.992 percent uptime for the processing of 20 million customers per day at more than 5,000 stores around the world.
The system maintains inventory data on more than 693 million items, of which 335 million are reviewed for potential reordering each day.
Wal-mart has been one of the leaders in adopting RFID, issuing a mandate that required its top 100 suppliers to start tagging all cases and pallets of merchandise by January 2005.
Wal-mart has estimated that RFID cut the incidence of out-of-stock products by 30 percent and improved the efficiency of moving products from backrooms to store shelves by 60 percent.

5 | *

Building a Responsive Organization/Supply Chain

Forecasting is always necessary for planning purposes, but when innovative products with unpredictable demand are ordered, accuracy tends to be greatly compromised.
When faced with an innovative product with unpredictable demand, a company should seek to make its supply chain more responsive by:

Deploying excess buffer capacity.

Deploying buffer stocks of parts or finished goods.

Investing in lead-time reduction.

Selecting suppliers on the basis of speed, flexibility, and quality rather than cost.

Employing a modular design.

demand

periods

last

total

...

188

225

228