Part 7-Forecasting PDF
Part 7-Forecasting PDF
Part 7
Dr. A NOORUL HAQ
Contents:
Causal Methods:
• Linear Regression
2
Causal Methods:
• Causal methods are most sophisticated forecasting tools.
• They are used when historical data are available and the
relationship between the factors to be forecasted from other
external or internal factors can be identified.
• These relationships are expressed in mathematical terms and can
be very complex.
• Linear regression is one of the causal methods.
3
Linear Regression
• This is one of the best known causal methods of forecasting.
• In this approach, one variable called dependent variable is related
to one or more independent variable by a linear equation.
• Dependent variables such as demand or cost is the variable we
want to forecast.
• Independent variables are assumed to affect the dependent
variable
4
• Thereby, caused (produced or induced or make) cause the results
observed in the past.
• Linear regression methods requires that a relationship between
the dependent variable and independent variable be built up first.
• Equation for regression line is Y = a + bX
where,Y is dependent variable and X is independent variable.
5
Y = numerical value of a point on the line for any given value of X.
a = numerical value ofY intercept of the line.
b= numerical value of slope of the line.
• The values for ‘a’ and ‘b’ are estimated from the sample data.
• The estimated value of a and b are obtained from the technique
called least squares.
• The equations used to determine these values are,
σ 𝑋 2 σ 𝑌 − σ 𝑋 σ 𝑋𝑌
a= 2
𝑛 σ 𝑋 2 −(σ 𝑋)
𝑛 σ 𝑋𝑌 − σ 𝑋 σ 𝑌
b= 2
𝑛 σ 𝑋 2 −(σ 𝑋)
6
• After calculating a and b, regression line can be plotted by
selecting 2 values of ‘X.’
• Calculating the value of ‘Y’ at these points and connecting the
points with a straight line.
• The two extreme values of ‘ X’ can be chosen for this purpose.
• The sample of data is limited to the two extreme values of ‘X’.
• There will not be any evidence to indicate the effects outside that
range.
• The relationship could even be non-linear.
7
• Consequently, making estimates outside the range of the data
would be very rare.
• Regression analysis can provide useful guidance for important
production management decisions.
• However, considerable amount of data is needed in order to obtain
useful relationship from linear regression models.
• This approach is relatively costly because of the amount of data
gathering and analysis is required to arrive at proper model.
8
Problem 1:
• Investigation of the demand of water coolers in 5 towns has
resulted in the following data.
Population of town (lakhs): 5 7 8 11 14
No. of coolers demanded: 45 65 55 75 95
• Fit a linear regression line and estimate the number of coolers
required for a town with a population of 12 lakhs.
X Y X2 XY
5 45 25 225
7 65 49 455
8 55 64 440
11 75 121 825
14 95 196 1330
45 335 455 3275 9
Solution 1:
𝑛 σ 𝑋𝑌 − σ 𝑋 σ 𝑌 5∗3275 −(45∗335)
b= 2 = = 5.20
𝑛 σ 𝑋 2 −(σ 𝑋) 5∗455 − (45)2
𝑛 σ 𝑋𝑌 − σ 𝑋 σ 𝑌 8∗4070 −(36∗865) 12
b= 2 = = 4.23
𝑛 σ 𝑋 2 −(σ 𝑋) 8∗204 − (36)2
Solution 2:
• The equation for the regression line is Y = a + bX
= 89.11+4.23X
• Forecasted demand for January, that is for X=9
Y= 89.11+4.23*9 = 127.18
• Forecasted demand for February, that is for X=10
Y= 89.11+4.23*10 = 131.41
• Forecasted demand for March, that is for X=11
Y= 89.11+4.23*11 = 135.64
13
END
14