THỐNG KÊ KINH DOANH_Ch10
-
Upload
thanh-tam-nguyen -
Category
Documents
-
view
219 -
download
0
Transcript of THỐNG KÊ KINH DOANH_Ch10
-
7/30/2019 THNG K KINH DOANH_Ch10
1/23
BUSINESS STATISTIC
NGUYEN TRUONG SON
-
7/30/2019 THNG K KINH DOANH_Ch10
2/23
CONTENTS
Chapter 1 Collecting data
Chapter 2 Presenting data Chapter 3 Summarizing data
Chapter 4 Correlation & Regression
Chapter 5 Probability & Expected value
-
7/30/2019 THNG K KINH DOANH_Ch10
3/23
CONTENTS (Cont.)
Chapter 6 The normal distribution and
confidence interval Chapter 7 Financial model
Chapter 8 Linear programming
Chapter 9 Index numbers
Chapter 10 Time based forecasting model
-
7/30/2019 THNG K KINH DOANH_Ch10
4/23
CHAPTER 10
Time based
forecastingmodel
-
7/30/2019 THNG K KINH DOANH_Ch10
5/23
LEARNING OBJECTIVE
To introduce the idea of time based models and their
uses in forecasting.
To discuss meaning of trend in data and the use of
linear regression to estimate them.
To develop the ideas of forecasting time series,
discussing component involve.
-
7/30/2019 THNG K KINH DOANH_Ch10
6/23
Introduction
-
7/30/2019 THNG K KINH DOANH_Ch10
7/23
10.1 Linear time based models
Data collected over a period of time is
usually called a time series. As discussed previously, this type of data is
normally display on a line graph with time
periods forming the x-axis and data values
shown on the y-axis
-
7/30/2019 THNG K KINH DOANH_Ch10
8/23
10.1 Linear time based models
0
10
20
30
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999No.oflicences
Year
No. of colour TV liences (millions)
Trends in time seriesTrend is usually used to describe the underlying direction
of a time series. In example 10.1 above, the data has an
upward trend as it clearly increases over time.
-
7/30/2019 THNG K KINH DOANH_Ch10
9/23
10.2 Forecasting using simple linear models
Once the linear model has been estimated,
the production of forecasts is relativelystraightforward.
However, it should be remembered that in
forecasting into the future extrapolation is
used. Assumption that the pattern seen in the past
will continue into the future.
-
7/30/2019 THNG K KINH DOANH_Ch10
10/23
10.3 Components of a time series
Trend (T)
As already discussed for the simple linearmodels, the trend is the underlying movement in
the data it may upward, downward or staionary.
Seasonal factors (S)
These are regular fluctuations within a completetime period (a day, a week, a month v.v.)
The important thing about seasonal factors is that
they represent some sort of repeating pattern
-
7/30/2019 THNG K KINH DOANH_Ch10
11/23
10.3 Components of a time series
Cyclical factors
These are long term fluctuations in the data andare similar to seasonal factors.
Its difficult to identify unless a long series of data
is available.
They may be related to economic factors exampleholiday sales may reduce during periods of
recession.
-
7/30/2019 THNG K KINH DOANH_Ch10
12/23
10.3 Components of a time series
Random or residual factors (S)
As in statistical analysis, there will be someunpredictable element to the data.
Example: Something like unusual weather
conditions effecting holiday sales.
The effects can not be forecasted.
-
7/30/2019 THNG K KINH DOANH_Ch10
13/23
10.3 Components of a time series
Since the cyclical elements will not consider
in detail here and random elements areimpossible to predict.
The model consists of isolating the two
predictable components:
Trend
Seasonal effects.
-
7/30/2019 THNG K KINH DOANH_Ch10
14/23
10.4 Forecasting the trend
The aim here is smooth out the effects due
to the other factors, leaving only theunderlying direction of the series to be
forecast.
This could be done simply be eye taking
some sort of straight line through the dataand attempting to the trend it.
-
7/30/2019 THNG K KINH DOANH_Ch10
15/23
10.4 Forecasting the trend
Moving average
Taking the average of each successive nobservations in turn formed by dropping the first
point in the sequence and picking up the next
one each time.
Average are generally take over the natural
periodof the data (the amount of time over which
the seasonal pattern reapeats)
For quarterly data is 4
For daily data is 5 or 7
-
7/30/2019 THNG K KINH DOANH_Ch10
16/23
10.4 Forecasting the trend
Moving average
Odd-point movingaverages
Day
No. of
enquires 5 point moving average
M 33
Tu 41W 77 66.2
Th 81 67.4
F 99 69
M 39 70.4
Tu 49 72.2
W 84 73.8Th 90 75.4
F 107 76.6
M 47 78
Tu 55 80.4
W 91 81.6
Th 102
F 113
-
7/30/2019 THNG K KINH DOANH_Ch10
17/23
10.4 Forecasting the trend
Moving average
Odd-point moving averages
0
20
40
60
80
100
120
M Tu W Th F M Tu W Th F M Tu W Th F
No. of enquires
5 point moving average
-
7/30/2019 THNG K KINH DOANH_Ch10
18/23
10.4 Forecasting the trend
Moving average
Even-point moving averagesYear Quarter Holiday sales 4-point moving average Centered moving average1996Q1 3576
Q2 2927 2,894.25
Q3 2710 2,865.75 2,880.00
Q4 2364 2,790.75 2,828.25
1997Q1 3462 2,692.00 2,741.38
Q2 2627 2,587.00 2,639.50Q3 2315 2,527.00 2,557.00
Q4 1944 2,550.00 2,538.50
1998Q1 3222 2,624.75 2,587.38
Q2 2719 2,649.00 2,636.88
Q3 2614
Q4 2041
-
7/30/2019 THNG K KINH DOANH_Ch10
19/23
10.4 Forecasting the trend
Moving average
Even-point moving averages
0
500
1000
1500
2000
2500
3000
3500
4000
Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4
1996 1997 1998
Holiday
Sales
Holiday sales data with centred moving average
Holiday sales
Centered moving average
-
7/30/2019 THNG K KINH DOANH_Ch10
20/23
10.4 Forecasting the trend
Forecasting the trend from moving average
T = a + bx Where:
T : centered moving average trend
X : the time period
b =nxy - xynx- (x)2 2
ya = -
bxn n
-
7/30/2019 THNG K KINH DOANH_Ch10
21/23
10.5 Forecasting the seasonal effects
The additive model
y = T + S +C + E Since the forecast of C = the cyclical effects will
not be considered here, so the model is:
y = T + S + E
S+ E = y - T
-
7/30/2019 THNG K KINH DOANH_Ch10
22/23
10.5 Forecasting the seasonal effects
The multiplicative model
y = T * S * C * E Since the forecast of C = the cyclical effects will
not be considered here, so the model is:
y = T * S * E
S * E = y / T
-
7/30/2019 THNG K KINH DOANH_Ch10
23/23
THE END
MANY THANKSFOR
YOUR ATTENTION