You are on page 1of 5

Time Series Analysis

Second degree equation

The second degree equation is


Yˆ= b0 + b1x+ b2x2
Where x = X – X if n is odd
x = 2(X – X)if n is even

To obtain the values of b0, b1 and b2 we have to solve following three


normal equations:
Y = nb0 + b2x2
xY = b1x2
x2Y = b0x2+ b2x4

06/21/2021 Prepared by: Pravat Uprety 1


Cyclical Variation
Percent of Trend
In this method the cyclical variation is computed
by dividing the actual value (Y) by the
corresponding trend value (Yˆ ) for each value in

the time series and multiplying this calculation


by 100.
Percent of trend = Yˆ X 100
Y
Base line = 100, variation is measured from 100.

06/21/2021 Prepared by: Pravat Uprety 2


Relative Cyclical Variation
Y  ˆ
Y
Relative cyclical residual = X100

Base line = 0
Variation is measured from the 0.

Note: Both techniques measure the same


variation from the trend line.

06/21/2021 Prepared by: Pravat Uprety 3


Seasonal variation
• To measure the seasonal impact we need to compute
seasonal index

• Steps

• Computation of 4-quarter moving total for the time series


• Computation of 4-quarter moving average
• Computation of centered 4-quarter moving average
• Computation of percentage of actual to moving average
value

06/21/2021 Prepared by: Pravat Uprety 4


• Computation of modified index by discarding the highest and lowest
values for each quarter and averaging the remaining values (if
information of three of less year is given then we do not discarding
the highest and lowest values)
 
• Computation of adjusting constant
The four quarterly indices should total 400. The adjusting constant is
computed by
Adjusting constant = 400/Sum of modified indices

Computation of seasonal index


Seasonal index = Modified Index X Adjusting Constant

06/21/2021 Prepared by: Pravat Uprety 5

You might also like