Scribd
Upload
9 views20 pages

Forecasting

The document discusses time-series regression for forecasting sales, emphasizing the use of the Least Squares Method to fit a trend line to historical data. It outlines the equation for the best-fitting line and introduces various accuracy measures for evaluating forecast performance, including Mean Absolute Deviation and Mean Square Error. Additionally, it compares different forecasting models, such as Naïve Forecasting, Moving Averages, and Simple Exponential Smoothing, highlighting their effectiveness in improving forecast accuracy.

Uploaded by

zariyahalexismontefalco
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
9 views20 pages

Forecasting

The document discusses time-series regression for forecasting sales, emphasizing the use of the Least Squares Method to fit a trend line to historical data. It outlines the equation for the best-fitting line and introduces various accuracy measures for evaluating forecast performance, including Mean Absolute Deviation and Mean Square Error. Additionally, it compares different forecasting models, such as Naïve Forecasting, Moving Averages, and Simple Exponential Smoothing, highlighting their effectiveness in improving forecast accuracy.

Uploaded by

zariyahalexismontefalco
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 20

Mathematics of

Finance
(Math 325)
FORECASTING
2. Time –Series Regression (Linear Trend)

- Regression in Statistics is a term used to describe the process of


estimating the relationship between two variables (ex. time and sales)
-The relationship is estimated by fitting a trend line to past data and then
project to the future.
- The best fitting line could be eyeballed, but precisely it can be found
using Least Squares Method equation minimizing the sum of the squares
of the errors between the points on the line and the actual sales

Scatter and Trend Line


100

90

80

70

60
Sales

50

40

30

20

10

0
1 2 3 4 5 6 7 8 9 10 11 12 13
Year

Sales(in millions) Linear (Sales(in millions))


Problem
The owner of the Computerworld store needs to
forecast sales one year ahead to help decide how
much he needs to borrow from the bank to finance
his inventories. Data are provided below.
Scatter and Trend Line
100
90
80
70
60
Sales

50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13
Year
Sales(in millions) Linear (Sales(in millions))
The equation of the best-fitted straight line is:

𝑭𝒕 =a + bt
Where 𝑭𝒕 = estimated or forecast value of sales for t
a = intercept, or the point at which the trend
line intercepts the x-axis (sales)
b = slope of the trend line, or the rate of change
in sales
t = time, in this case year

a= 𝒙
ഥ - b𝑡 ҧ where 𝒙
ഥ= mean of the values of x
ഥ𝑡 = mean of the values of t

ҧ𝒙
(σ 𝒕𝒙 )−𝒏𝒕ഥ
b= σ 𝟐 ҧ𝟐 where x =dependent variable, sales
𝒕 −𝒏𝒕
𝒕ҧ = mean of the values of t
ഥ = mean of the values of x
𝒙
σ 𝒕𝒙 −𝒏𝒕ҧ 𝒙ഥ
b= σ 𝟐 ҧ𝟐
𝒕 −𝒏𝒕
6090− (12)(6.5)(72.4)
= = 3.0965
650− (12)(42.25)

ഥ - b 𝒕ҧ
a= 𝒙
= 72.4 – (3.0965) (6.5)
= 52.27275
𝑭𝒕 = a + bt
= 52.27275 + (3.0965) (13)
= 92.48484848 M
Sales(in millions)
100

90

80

70

60

50

40

30

20

10

0
1 2 3 4 5 6 7 8 9 10 11 12 13

Sales(in millions) Linear (Sales(in millions))


EVALUATING FORECAST ACCURACY

Accuracy Measures are:


1. Mean Absolute Deviation (MAD)--
has equal weight to each error
2. Mean Absolute Percentage Error
(MAPE)
3. Mean Square Error (MSE)-- gives
more weight to large errors because
they are squared
Note : It is up to the manager not the
management scientist, to decide which
forecast accuracy measure is most
appropriate for his or her application.

The MSE is most often used in practice.


So how do we know when our forecasts are
good, bad or indifferent?

One way to answer this question is to


compare the accuracy of a given model
with that of a benchmark model
Benchmark Model

Naïve Forecasting Equation: Ft+1 = X 𝑡


This means:
Forecast for the next period = Observed value
this period
Forecast Error Equation: 𝑒𝑡 = 𝑋𝑡 - 𝐹𝑡
This means,
Error = Data- Forecast
Warmed
up
samples
3rd Forecasting Model
MOVING AVERAGES
Consider the data on airport passenger
embarkations by month:
The Naïve Forecast of the given data is
as follows:
In unweighted moving average method, the forecast is the
mean of the last N data points. If N=3, the forecasting equation
is:

So from the data,

At the end of period 4, the oldest observation is dropped and


the observation for period 4 is added. The forecast for period 5
is

and the other forecasts follow in the same manner…


Table for a 3-Period Moving Average

Note that the MSE for moving average is better than that of the Naïve
benchmark, and this method is also promising.
4. Simple Exponential Method
In this method, the new forecast is equal to the old forecast plus a
fraction of the error. The fraction 𝛼 (smoothing parameter) lies
between 0 and 1.

In the given data, as a rule of thumb, let’s use 𝐹1 = 30 (the mean of the
warm-up sample) and 𝛼 = .1 (the least value). So ,
Note again that the MSE in exponential smoothing is an improvement of
the naïve model as well as that of the moving average method.

You might also like