Forecast values are constant

by Satish Karivedha   Last Updated February 13, 2018 12:19 PM

I ran a rainfall forecasting time series model using auto arima. P,Q,D values as per auto arima are (0,0,0). The forecast values for the next 5 years are constant. Enclosed below is the detailed steps for the same. Need a Clarification on this. In the forecasted values I could be derive 80% degree of freedom range and 95% degree of freedom of range. My theory on this is the actuals values may fall in between the degree of freedom range. Need some confirmation on the same

KA <- read.csv("C:/Users/Satish Karivedha/Desktop/KA.csv") head(KA) State Year Annual 1 KARNATAKA 1950 5429 2 KARNATAKA 1951 4758 3 KARNATAKA 1952 4397 4 KARNATAKA 1953 5610 5 KARNATAKA 1954 5536 6 KARNATAKA 1955 5522 names(KA) <-c("State", "Year","Rainfall") head(KA) State Year Rainfall 1 KARNATAKA 1950 5429 2 KARNATAKA 1951 4758 3 KARNATAKA 1952 4397 4 KARNATAKA 1953 5610 5 KARNATAKA 1954 5536 6 KARNATAKA 1955 5522 KA<- KA[order(KA$Year),] > head(KA) State Year Rainfall 1 KARNATAKA 1950 5429 2 KARNATAKA 1951 4758 3 KARNATAKA 1952 4397 4 KARNATAKA 1953 5610 5 KARNATAKA 1954 5536 6 KARNATAKA 1955 5522 > plot(Rainfall~Year,data=KA) KA<-ts(KA$Rainfall,start=min(KA$Year),end=max(KA$Year)) KA Time Series: Start = 1950 End = 2015 Frequency = 1 [1] 5429 4758 4397 5610 5536 5522 6139 5077 5649 6310 5186 7712 6164 4947 5210 4199 4711 [18] 4768 5189 5179 5907 4868 4420 4963 5302 6728 4256 5437 5999 4907 5679 5911 5776 5831 [35] 5041 4376 4585 4398 5284 4805 5688 5661 5752 5251 6437 4903 5045 6229 5674 5590 5271 [52] 4636 4079 4308 4845 5598 5518 5928 5059 5934 5780 5651 1462 6063 5625 4770 KAOPT<-auto.arima(KA) KAOPT Series: KA ARIMA(0,0,0) with non-zero mean

Coefficients: mean 5286.697 s.e. 100.272

sigma^2 estimated as 673803: log likelihood=-536.03 AIC=1076.06 AICc=1076.25 BIC=1080.44

predict(KAOPT,n.ahead=5,se.fit = T) $pred Time Series: Start = 2016 End = 2020 Frequency = 1 [1] 5286.697 5286.697 5286.697 5286.697 5286.697

$se Time Series: Start = 2016 End = 2020 Frequency = 1 [1] 820.8552 820.8552 820.8552 820.8552 820.8552

RainfallForecast<-forecast(object = KAOPT,h=5) RainfallForecast Point Forecast Lo 80 Hi 80 Lo 95 Hi 95 2016 5286.697 4234.729 6338.665 3677.85 6895.544 2017 5286.697 4234.729 6338.665 3677.85 6895.544 2018 5286.697 4234.729 6338.665 3677.85 6895.544 2019 5286.697 4234.729 6338.665 3677.85 6895.544 2020 5286.697 4234.729 6338.665 3677.85 6895.544

Tags : forecasting


Related Questions





Best way to deal with forecasting with noisy data?

Updated August 07, 2015 14:08 PM

Forecasting future returns ARMA-GARCH

Updated June 12, 2016 08:08 AM