When managers plan, they determine in present
what courses of action they will take in future.
First step in planning is FORECASTING i.e.
estimating the future demand for products and
services and the resources necessary to produce
Estimates of the future demand for products and
services are commonly called as sales forecasts.
Long-range forecasts used to make strategic
decisions about products, processes and
Short-range forecasts needed to assist in
decision making about operations issues that
span only the next few days or weeks.
Medium range forecasts group products into
Why forecasting is essential in
New facility planning
Qualitative Forecasting Methods
Based on the judgements about the causal
factors that underlie the sales of particular
products or services and on opinions about the
relative likelihood of those causal factors being
present in the future.
These methods may involve several levels of
Educated guess-when one person uses his or her
intution and experience to estimate a forecast.
Executive committee consensus and Delphi method
describe procedures for assimilating information
within a committee for the purpose of generating a
sales forecast and are useful for either existing or
new products and services.
Survey of sales force and survey of customers are
methods primarily used for existing products and
Historical analogy and market surveys describe
procedures that are useful for new products and
Forecasting method that is appropriate depends on
product’s life cycle stage.
Quantitative Forecasting Models
Mathematical models based on historical data.
Such models assume that past data are relevant
to the future.
These models can be used with time series.
Time series is a set of observed values measured
over successive time periods such as monthly
sales for the last two years.
A model uses least-square method to identify the
relationship between a dependent variable and
one or more independent variables that are
present in a set of historical observations.
In simple regression there is only one
In multiple regression there is more than one
If the historical data set is a time series, the
independent variable is the time period and the
dependent variable in sales forecasting is sales.
A regression model does not have to be based on
a time series.
In that case, the knowledge of future values of the
independent variable known as causal variables
is used to predict future values of the
Linear regression is ordinarily used in long range
Short range time series type of forecasting model
that forecasts sales for the next time period.
In this model the arithmetic average of the actual
sales for a specific number of most recent past
time periods is the forecast for the next time
Short range forecasting model that forecasts
sales for the next time period.
Forecast sales for the last period is modified by
information about the forecast error of the last
This modification of the last period’s forecast is
the forecast for the next time period.
Exponential smoothing with trend
The exponential smoothing model but modified to
accommodate data with a trend pattern.
Such patterns can be present in medium range
Also called double exponential smoothing
How close forecasts come to actual data.
Because forecasts are made before the actual
data become known, the accuracy of forecasts
can be determined only after the passage of time.
If forecasts are very close to the actual data we
ssay that they have high accuracy and forecast
error is low.
Long range forecasts
Estimating future conditions over time spans that
are usually greater than one year.
Historical data is made of several components.
These components are cycles,trends,seasonality
and random fluctuations or noise.
Long range trends are upward or downward
Cycle is a data pattern that may cover several
years before it repeats itself again.
Noise is a pattern resulting from random variation.
Seasonality is a pattern that repeats itself after a
period of time usually one year.
Short range forecasts
Estimates future conditions over time spans that
range from a few days to several weeks.
These forecasts may span such short periods of time
that cycles,seasonality and trend patterns have little
The main data pattern affecting these forecasts is
These forecasts provide information to make such
decisions as :
How much inventory of particular product be carried next
How much of each product should be scheduled for
production next week?
Evaluating forecasting model
Impulse response:-forecats that respond very fast
to changes in historical data have high impulse
When forecasts reflect few of the changes in
historical data, these forecasts are said to have
low impulse response.
Noise Dampening Ability:-forecasts that reflect
every little happenstance fluctuation in the past
data are said to include random variation or
Such forecasts are unpredictable from period to
It is usually desirable to have short range
forecasts that have both high impulse response
and high noise dampening ability but this is not
A forecasting system that responds very fast to
changes in the data has a great deal of random
Forecasters must choose which characteristic has