| dc.description.abstract | —This study mainly focuses on developing a
statistical model to predict ASPI with a reasonable
accuracy and identifying the relationships between ASPI
and other important socio-economic variables, such as
macroeconomic variables, elections and war. The key
results of this study include that the absence of the war has
a positive relationship with ASPI compared to the presence
of the war and the presence of severe bomb blasts in
Colombo. The presence of presidential and general
elections and the crude oil price have a positive
relationship with ASPI. Furthermore, presence of the
provincial elections, Imports, exports and exchange rates
have a negative relationship with ASPI. Moreover, it was
found that during the post war period the conditional
heteroscedasticity was absent in ASPI. In addition to that
the model which was chosen to predict the ASPI has a
Mean Absolute Percentage error (MAPE) of 4.05% on the
test set | |
