Page 45 - ASBIRES-2017_Preceedings

Kaushalya & Francisco

H :series is stationary The Phillips-Perron Test was performed
1
Augmented Dickey-Fuller Test to identify the stationarity of the difference
series.
data: Dengue_Kurunegala_No.of. cases Phillips-Perron Unit Root Test
Dickey-Fuller = -2.0651, Lag order = 4,
p-value = 0.5491 data: non_seasonal_difference
alternative hypothesis: stationary Dickey-Fuller Z(alpha) = -56.34, Truncation
lag parameter = 3, p-value = 0.01
Since p value (0.5491) was greater than
the 5% significance level. The result alternative hypothesis: stationary
indicated, the series is not stationary at 5% Warning message:
significance level. Non-seasonal
differencing was done for the original data. Inpp.test(non_seasonal_difference): p-value
smaller than printed p-value

Figure 2: Autocorrelation function of
original data

Figure 5: Partial autocorrelation function of
seasonally differenced data

The Phillip Perron test concluded that
the seasonally differenced series is
stationary at 5% significance level.
Figure 3: Autocorrelation function of non- The given below summary is about the
seasonal differenced data
cut off lags of seasonal and non-seasonal
Phillips-Perron Test confirmed that lags.
the differenced data were stationary.A
specific pattern with equal length of six was Table 1: Behavior of ACF and PACF
in the graph which indicated there was a
seasonal pattern. Therefore, a seasonal Seasonal Non-
difference of length six was done. seasonal

ACF Cuts off at Dies down
lag 1 quickly
Cuts off Dies down
PACF
lag 1 quickly

Differences 1 1

Figure 4: Autocorrelation function of
seasonally differenced data

35