FORECASTING

TECHNQUES
• Since planning and budgeting involve looking into the future, the
company must make some assumptions about the outlook for the
environment in which its business operates. These assumptions are
called premises.
• When identifying premises, it is essential for management to focus
only on those premises that will actually impact the potential success
of the business. Focusing on premises that are not a critical part of
the organization’s success wastes valuable time and resources.
• MATHEMATICAL MODEL
A mathematical model is an equation that attempts to represent an
actual situation.
LINEAR REGRESSION ANALYSIS
• Linear regression is used to predict the value of a variable based on
value of another variable. The value you want to predict is dependent
variable and the variable used to predict is independent variable
• Relationship between independent and dependent variable will be
linear(straight line)
TWO METHODS ARE USED
• 1) Time series methods, which look only at the historical pattern of
one variable and generate a forecast by extrapolating the pattern
using one or more of the components (or patterns) of the time series,
• 2) Causal forecasting methods, which look for a cause-and-effect
relationship between the variable being forecasted (the dependent
variable) and one or more other variables (the independent
variables).
TIME SERIES ANALYSIS
• A time series is a sequence of measurements taken at equally spaced,
ordered points in time. In a time series, the causal factor is the
passage of time. Time series measurements reflect historical activity
for one variable over a sequence of past time periods.
• A time series method of forecasting involves looking for a pattern in
the variable’s historical values that can be used to forecast the future.
• Here independent variable is time.
• TREND PROJECTION
Using historical data to forecast values for future.
LINE OF BEST FIT
• A line of best fit is a straight line that minimizes the distance between
it and some data. The line of best fit is used to express relationship
between in a scatter plot of different data points. It is an output of
regression analysis and can be used as a prediction tool
• The difference between each point and its corresponding point on the
regression line is called a deviation or a residual.
• The regression line is called a trend line when the regression is being
performed on a time series with a trend pattern.
CAUSAL FORECASTING
• Causal forecasting methods are used when the value being forecast
(the dependent variable) is affected by some other value or values
(the independent variable or variables). If a cause-and-effect
relationship can be identified between the independent variable or
variables and the dependent variable and if that relationship is a
linear one, a projection of the independent variable or variables can
be used to forecast the dependent variable.
MULTIPLE LINEAR REGRESSION
ANALYSIS
• If only one independent variable (such as advertising expenditures) is
affecting the dependent variable (such as sales revenue) and if the
relationship between them is linear, the regression analysis is simple linear
regression. However, it is also possible for a dependent variable such as
sales revenue to be affected by more than one independent variable. For
example, advertising expenditures, the size of the sales staff, competition,
the economy, or any number of other circumstances can impact sales
revenues. When more than one independent variable is known to impact
the dependent variable and each one can be expressed numerically,
regression analysis using all the independent variables to forecast the
dependent variable is called multiple regression analysis. Multiple
regression analysis is a type of causal forecasting.
• GOODNESS OF FIT
The goodness of fit measures the reliability of the regression. R-
squared is the quantitative measurement of the degree that a change
in the dependent variable is predicted by the change in the
independent variable. This number will be between zero and 1. The
closer the number is to 1, the more reliable the regression is.
• 2. Statistical Reliability of Each Independent Variable
Another quantitative measure of reliability is the statistical reliability of
each independent variable. This measure is called the t-value. It
measures if an independent variable (X) has a valid, long-term
relationship to the dependent variable (y). A good rule of thumb is that
the t-value should be greater than 2. A lower number indicates that
there is little or no statistical relationship between the variables
• Standard error of estimate
The standard error of estimate (SE) measures the accuracy of y, the
regression’s estimate. It is a measurement of the dispersion around the
regression line. A relatively small SE is better than a large one. It implies
that the actual value will lie in the range of the estimate of y, plus or
minus the SE
LEARNING CURVE
• The term “learning curve” refers to the concept that efficiency
increases as the amount of experience a person has with a given task
increases. As a result, the time required for performing the task
decreases as increases occur in the number of times the task has
been performed.
• Learning curve analysis is used in planning, budgeting, and forecasting
and also to determine estimated labor costs when bidding on a
contract. A company needs to be able to estimate what the long-term
costs of production will be.
• Two learning-curve models are commonly used:
• 1) The cumulative average-time learning model is based on the assumption
that the cumulative average time required per unit declines at a constant rate
each time the cumulative quantity of units produced doubles. The cumulative
average-time learning model can be used to estimate the average time per
unit required to produce all of a given number of units produced.
• 2) The incremental unit-time learning model is based on the assumption that
the incremental amount of time required to produce the last unit declines at a
constant rate each time the cumulative quantity of units produced doubles.
The incremental unit-time learning model can be used to estimate the time
needed to produce the last unit in a quantity of units.
• Learning curve should be below 100%
• Learning curve should be more than 50%
• If it is below 50% the total time required to produce additional units
plus time required to produce initial units will be less than time
required to produce initial units
• If it is 50% the total time required to produce additional units plus
time required to produce initial units will be equal to the time
required to produce initial units
CUMULATIVE AVERAGE TIME
LEARNING MODEL
• The cumulative average-time learning model assumes a constant rate
of decline in the estimated cumulative average time per unit each time
the quantity of units produced doubles.
• It can be used to calculate:
1. Estimated average time for entire unit produced
2. Total time required to produce the entire quantity produced
3. The total estimated time required to produce certain block of units
(finding total time required for units through the end of that block
and deducting from that total time required for units up to that
block)