FORECASTI

NG
 A planning tool that helps
management in its attempt
to cope with uncertainties
 Enables manager to
anticipate the future so they
can plan accordingly
 2 Uses of Forecasting

1. Plan the system

- involves long range of plans about
the type of product, services to offer,
facilities to have and where to put.

2. Plan the use of the system

- refers to short- range of
planning which involve tasks.
 Features Common to All
Forecasts
Forecasting techniques generally assume that
the same underlying casual system that exited in
the past will continue to exist in the future.
Forecast rarely perfect
Forecast for groups of items tend to be more
accurate than forecast for individual items.
Forecastaccuracy decreases as the time, period
covered by the forecast increases.
Elements of a Good Forecast

The forecast should be:

Timely
Accurate
Reliable
Meaningful units
Writing
Simple to understand and use
Cost effective
Steps in Forecasting Process

Determine the purpose of the

forecast
Establish a time horizon
Select a forecasting technique
Obtainclean, and analyze
appropriate data
Make the forecast
Monitor the forecast
 Approaches to Forecasting

Qualitative Quantitative
techniques techniques

- Based on opinion - Used to forecast

or judgment future data as a
function of past
data
 Variety of Forecasting
Techniques

Judgmental forecast
Time series forecast
Associative models
 Variety of Forecasting
Techniques
a. Executive Opinion
- directors of various operational
function gather together to generate their
own predictions on prospective sales
estimates.

b. Sales force Opinion

- often good sources of information
because of their direct contact with the
customer.
 Judgmental Forecast

c. Consumer Survey
- this method removes any internal bias
within the company from the forecast results.

d. Delphi Method
- series of questionnaires submitted to
panel of industry experts.
 Variety of Forecasting
Techniques
Time series Associative
forecast models

- Project patterns - Technique that uses

identified in explanatory
recent time-series variables to
observation predict future
demand.
Forecast based on time series

Time series- a time ordered sequence of

observation taken at regular intervals.
Trend-
a long term upward or downward
movement in data.
Seasonality- short term regular variation
related to the calendar or time of day
Cycles-wavelike variation of lasting more than
one year.
Forecast based on time series

Irregularvariation- cause by unusual

circumstances not reflective of typical
behavior
Techniques for averaging
Moving average- technique that averages a
number of recent actual values, updated as new
values become available.
Weighted moving average- recent values in a
series are given more weight in completing a
forecast.
Exponentialsmoothing- a weighted average
method based on previous forecast lost a
percentage of the forecast error
Techniques for trend

Lineartrend equation- F1 = a + bt, used to

develop forecasts when trend is present
Trend- adjusted exponential
smoothing
Variation of exponential smoothing-
used when a time series exhibits a
linear trend.
Formula:

TAFt+1 = St +Tt
Techniques for seasonal variation

Seasonal variations- regularly

repeating movements in series
values that can be tied to recurring
events.
Techniques for Cycles

Cycles- up and down movements

similar to seasonal variations but of
longer durations
 Seasonality expressed in terms of the
amount of that actual value that deviate
from the average value of series.

2 Modes of Seasonality:
Additive
Multiplicative
 SeasonalRelative seasonal
percentages in the multiplicative
model.

1. Deseasonalize Data
2. Incorporate Seasonality
Centered Moving Average

Commonly used method for

representing the trend portion of a
time series.
Cycles

Upand down movements similar to

seasonal variation but of longer durations.
Regression

Primary method of analysis.

Correlation

Measures the strength and direction

of relationship between the two.
Associative forecasting techniques

Predictor variables- can be used to predict

values of the variable interest.
Regression- technique for fitting a line to a set
of points.
Leastsquares line- minimizes the sum of the
squared vertical deviations around a line.
Yc= a + bx
USE OF LINEAR REGRESSION
ANALYSIS
Assumptions that have been satisfied:
Variations around the linear is
random.
Deviationsaround the linear should
normally be distributed.
Predictionsare made within a range
of observed values.
USE OF LINEAR REGRESSION
ANALYSIS
Always plot the data to verify that a linear
relationship is appropriate.
Incase when data is time dependent, plot the
dependent variable versus time; if pattern
appears, use analysis of time series instead of
regression, or use time as an independent
variable as part of a mulitiple regression
analysis.
A small correlation may imply that other
variables are important.
USE OF LINEAR REGRESSION
ANALYSIS
Weaknesses of regression:
Simple linear regression applies only to linear
relationships with one independent variable.
Needs a considerable amount of data to
establish the relationship- in practice, 20 or
more
All observations are weighted equally.
EXAMPLE:
Sales of new houses and three-month lagged
unemployment. Determine if unemployment levels can
be used to predict demand for new houses.
Period Unemployment % (x) Units Sold (y)
1 7.2 20
2 4.0 41
3 7.3 17
4 5.5 35
5 6.8 25
6 6.0 31
7 5.4 38
8 3.6 50
9 8.4 5
10 7.0 19
11 9.0 14
 SALES OF NEW HOUSES AND THREE-MONTH
LAGGED UNEMPLOYMENT
60

50

40
Units Sold (y)

30

20

10

0
3 4 5 6 7 8 9 10
Level of Unemployment (x)
CORRELATION COEFFICIENT

r= -9.66 high negative correlation

REGRESSION EQUATION

y= 71.85 6.9x
CURVILINEAR AND MULTIPLE
REGRESSION OF ANALYSIS

A model that involves more than one

predictor require the use of multiple
regression analysis.
ACCURACY AND CONTROL OF
FORECASTS
Forecast errors difference between
the value that occurs and the value
that was predicted for a given
period of time. Hence,

e t = A t - Ft
MEASURES FOR SUMMARIZING
HISTORICAL ERRORS
Mean Absolute Deviation (MAD)-
the average absolute forecast
error.

MAD=
MEASURES FOR SUMMARIZING
HISTORICAL ERRORS
Mean Squared Error (MSE)- the
average of squared forecast
errors.

MSE=
MEASURES FOR SUMMARIZING
HISTORICAL ERRORS
MeanAbsolute Percent Error
(MAPE)- the average absolute
percent error

MAPE=
Example:
Compute the MAD, MSE, and MAPE for the ff. data,
showing actual and predicted number of accounts service.
PERIOD ACTUAL FORECAST (A-F) |ERROR| ERROR 2
(|ERROR\
ERRO ACTUAL) X
R 100
1 217 215 2 2 4 .92%
2 213 216 -3 3 9 1.41%
3 216 215 1 1 1 .46%
4 210 214 -4 4 16 1.9%
5 213 211 2 2 4 .94%
6 219 214 5 5 25 2.28%
7 216 217 -1 1 1 .46%
8 212 216 -4 4 16 1.89%
-2 22 76 10.96%
SOLUTION:

MAD= =22\8= 2.75

MSE= =76\(8-1)= 10.86

MAPE= = 10.26%\ 8= 1.28%

Controlling the
Forecast
Variety of possible sources of forecast
errors:

The model may be inadequate due to:

a. the omission of an important variable
b. a change or shift in the variable that the model cannot deal with
c. the appearance of a new variable
Irregular variations may occur due to severe weather or other natural
phenomena, temporary shortage or breakdowns, catastrophes, or similar
events.
The forecasting techniques may be used incorrectly, or the results
misinterpreted.
There are always random variations in the data.
Control chart

A visual tool for monitoring forecast

errors.
 upper control limit

range of
error of zero
random
variability

lower control limit

Tracking signal

Theratio of cumulative forecast

error to the corresponding value of
MAD, use to monitor a forecast.
Choosing a forecasting
technique
Two most important factors:

Cost
Accuracy
Other factors

The availability of historical data

Theavailability of computer
software
Timeneeded to gather and analyze
data and to prepare the forecast
Using forecast information

Reactive Approach
-it views forecasts as probable future demand ,
and a manager react to meet that demand.
Proactive Approach
- it seeks to actively influence demand. it requires
either an explanatory model or a subjective
assessment of the influence on demand.
Forecasting Amount of Data Pattern Forecast Preparation Personnel Background
Method Historical Data Horizon Time

Moving average 2-30 Data should be short short little sophistication

observations stationary

Simple 5-10 Data should be short short little sohistication

exponential observations stationary
smoothing

Trend-adjusted 10-15 Trend short to medium short moderate sophistication

exponential observations
smoothing

Trend models 10-20; for Trend short to medium short moderate sophistication
seasonality at
least 5 per
season

Seasonal enough to see 2 Handles cyclical short to medium short to little sophistication
peaks and and seasonal moderate
troughs patterns

Causal 10 observations Can handle short, medium, long considerable

regression per independent complex or long development sophistication
models variable patterns time, short time
for
implementation
Factor Short range Intermediate Long range
range
1. Frequency often occasional infrequent
2. Level of item productive total output
aggregation family type of
product/service

3. Type of smoothing projection managerial

model projection seasonal judgement
regression regression

4. Degree of low moderate high

management
involvement

5. Cost per low moderate high

forecast