Professional Documents
Culture Documents
NOVEMBER 2020
TSA ASSIGNMENT
on
“FORECASTING”
SUBMITTED TO:
DR. PANKAJ KR. VARSHNEY
SUBMITTED BY:
SHILPI MURARKA (420 / 2019)
dlnBelgium
At first difference, series seems to have a constant mean, and now we can check AR and MA
processes.
So, we take correlogram of first difference
We can see from Correlogram that ACF and PACF are significant only for 1 level only, so we
must check ARMA levels for those ARIMA (p,d,q) models :- (0,1,1), (1,1,0),(1,1,1)
ARMA Check at (1,1,0) levels
As we find that AIC is highest for ARIMA(1,1,1) model we will prefer ARIMA (1,1,1) model
for further forecasts and analysis.
Now we look for coefficients of AR(1), MA(1) and Constant using Conditional
Least Squares
We have taken data till 2014Q4 as sample and used Forecast function on ARIMA(1,1,1) model
to predict remaining 8 quarters of data and the above picture shows E-Vies Forecast result.