You are on page 1of 2

# OM7 6.

Chapter 8: Forecasting exponential smoothing model that is suited to data
that exhibit a trend.
Forecasting - predicting future events. 7.Linear Trend Line computes a forecast with
trend by drawing a straight line through a set of
Principles of Forecasting: data.
1. Forecasts are rarely perfect. 8. Seasonal Indexes percentage amount by
2. Forecasts are more accurate for groups or which data for each season are above or below the
families of items rather than for individual items. mean.
3. Forecasts are more accurate for shorter than
longer time horizons. Casual (Associative) Models:
1. Linear Regression procedure that models a
Steps in the Forecasting Method: straight-line relationship between two variables.
1. Decide what to forecast. 2. Multiple Regression similar to linear but
2. Evaluate and analyze appropriate data. models the relationship of multiple variables with
3. Select and test the forecasting model. the variable being forecast.
4. Generate the forecast.
5. Monitor forecast accuracy. Four Basic Patterns in a Time Series Model:
1. Level or Horizontal Pattern a pattern in
Qualitative Forecasting Methods forecast is which data values fluctuate around a constant
made subjectively by the forecaster. mean.
2. Trend a pattern in which data exhibit increasing
Quantitative Forecasting Methods forecast is or decreasing values overtime.
based on mathematical modeling. 3. Seasonality any pattern that regularly repeats
itself and is constant in length.
Types of Forecasting Methods: 4. Cycles data patterns created by economic
1. Qualitative Forecasting Methods fluctuations.
-Characteristics: based on human judgement,
opinions, subjective and non-mathematical Random Variation is an unexplained variation
-Strengths: can incorporate latest changes in the that cannot be predicted.
environment and inside information Data= level + trend + seasonality + cycle +
-Weaknesses: can bias the forecast and reduce random variation
forecast accuracy Data= pattern + random variation
2. Quantitative Forecasting Methods
-Characteristics: based on mathematics, Correlation Coefficient is statistic that measures
quantitative in nature the direction and strength of the linear relationship
-Strengths: consistent and objective, able to between two variables.
consider much information and data at one time
-Weaknesses: often quantifiable data are not Forecast Error is the difference between the
available, only as good as the data on which they forecast and actual value for a given period.
are based Et= At Ft
Et= forecast error for period t
Qualitative Forecasting Methods: At= actual value for period t
1. Executive Opinion forecasting method in Ft= forecast for period t
which a group of managers collectively develop a
forecast. Most Commonly Used Error Measures:
2. Market Research approach to forecasting that 1. Mean Absolute Deviation (MAD) measure of
relies in surveys and interviews to determine forecast error that computers error as the average
customer preferences. of the sum of the absolute errors.
3. Delphi Method approach to forecasting in MAD= |actual forecast|/n
which a forecast is the product of a consensus 2. Mean Squared Error (MSE) measure of
among a group of experts. forecast error that computers error as the average
of the squared error.
Quantitative Forecasting Methods: MSE= (actual forecast)2/n
1.Time Series Models based on the assumption
that a forecast can be generated from the Forecast Bias a persistent tendency for a
information contained in a time series of data. forecast to be over or under the actual value of the
2.Time Series a series of observations taken over data.
time.
3.Casual Models based on the assumption that Tracking Signal tool used to monitor the quality
the variable being forecast is related to other of a forecast.
variables in the environment. Tracking Signal= algebraic sum of forecast
Time Series Models: Tracking Signal= (actual forecast)/MAD
1. Nave Method forecasting method that
assumes next periods forecast is equal to the Factors in Selecting the Right Forecasting
current periods actual value. Model:
2. Simple Mean Average the average of a set of 1. Amount and type of available data.
data. 2. Degree of accuracy required.
3. Simple Moving Average a forecasting method 3. Length of forecast horizon.
in which only n of the most recent observations are 4. Data patterns present.
average.
4. Weighted Moving Average a forecasting Forecasting Software:
method in which n of the most recent observations 1. Spreadsheets
are averaged and past observations may be 2. Statistical Packages
weighed differently. 3. Specialty Forecasting Packages
5. Exponential Smoothing uses a sophisticated
weighted average procedure to generate a forecast.
Guidelines for Selecting Forecasting Software: between two trading partners that establishes
1. Does the package have the facilities you want? formal guidelines for joint forecasting and planning.
2. What platform is the package available for?
3. How easy is the package to learn and use? Nine Steps to Utilize CPFR:
4. It is possible to implement new methods? 1. Establish collaborative relationships.
5. Do you require interactive or repetitive 2. Create a joint business plan.
forecasting? 3. Create a sales forecast.
6. Do you have large data sets? 4. Identify exceptions for sales forecasts.
7. Is there any local support? 5. Resolve/collaborate on exceptions to sales
8. Does the package give the right answers? forecasts.
6. Create order forecast.
Focus Forecasting is a forecasting approach that 7. Identify exceptions for order forecast.
has gained some popularity in business. 8. Resolve/collaborate on exceptions to order
forecast.
Collaborative, Planning, Forecasting and 9. Generate order
Replenishment (CPFR) is a collaborative process