![](bg2.png)
(Mustofa, 2010). Muzakir forecasted the magnitude
of earthquake using Neural Networks in time series
data (Sultan, 2014). Jordan Grestandhi et.al analyzed
the OLS-ARCH and ARIMA method to predict the
stock price index (Jordan Grestandhi et al, 2011).
Bayu Ariestya Ramadhan analyzed the comparison of
ARIMA and GARCH methods to forecast stock
prices (Ramadhan, 2015). Ahmad Sadeq forecasted
stock index using ARIMA method (Sadeq, 2008).
The results of this forecasting are accurate with
MAPE 4.14%. Chintya Kusumadewi forecasted the
Indonesia stock market price index using the ARIMA
and Genetic Programming (Kusumadewi, 2014). In
this research, the MAPE value is smaller compared to
using the ARIMA method only, which is
1.181192667%. Ping-Feng Pai and Chih-Sheng Lin
combined ARIMA and Support Vector Mechine
(SVM) methods to forecast stock price (Ping-Feng
Pai and Chih-Sheng Lin, 2005). Nurissaidah
Ulinnuha and Yuniar Farida forecasted the weather
using ARIMA – Kalman Filter (Ulinnuha & Farida,
2018). In this research, it is known that the ARIMA
kalman filter method is optimal for forecasting, with
MAPE value of 0.000389.
From several studies above, most of the ANN
methods are proven to be more optimal when
compared to statistical methods such as ARIMA. This
is because ARIMA models cannot identify nonlinear
patterns of data (Shukur, 2015). So, the results of
forecasting by ARIMA model must be increased
using other methods, such as Kalman Filter. The
Kalman Filter approah is used as an optimal solution
for many data tracking and predictions, because the
Kalman filter reduces noise and obtains correct data
(Hairong Wang et al, 2017). So, in this research, the
results of ARIMA forecasting were improved by
using Kalman Filter to obtain a more optimal
forecasting result and compare it with the ANN
method.
2. THEORITICAL FRAMEWORK
2.1 Indonesia Sharia Stock of Index (ISSI)
ISSI is index of stock that covers all sharia stock in
Indonesia and registered in Indonesia Stock
Exchange. The difference between ISSI and stock in
general is that the implementation of ISSI does not
violate religion. Stocks of Sharia have some criteria.
They are:
1. Activities carried do not violate Islamic religious
law. They are :
a) Everything that belongs to gambling
b) Trade is prohibited sharia
c) Based financial services of usury (riba)
d) Traded risks contain uncertain elements or
gambling
e) Producing, distributing, trading and providing
illicit goods or services that defined by DSN-
MUI
f) Transactions carried contain elements of
bribery
2. Confirm of financial ratios. They are :
a) Total money based on interest compared to
total assets not exceeding 45%
b) Total interest income and other haram income
compared to total business income and other
income not exceeding 10%
2.2 Time Series Analysis
Time series data are some data from a specific
variable successive in each period, for example daily,
weekly, monthly, yearly and etc. Time series data are
important to predict next occurrences. because it is
known that multiple data patterns of the past will be
repeated in the future. Any observations made can be
expressed in random variables
which is obtained
on a certain time index
, with
. so, from time series, data can be
written with
2.3 Stationary Test of Time Series Data
Stationary data is when the data pattern is at
equilibrium around the constant mean and the
variance around average which is is constant for some
time (S. Makridakis, S. C. Wheelwright, and V. E.
McGee, 1999). The data that are not stationary
against variance, it must be transformed by the Box-
Cox transformation method (G. E. P. Box, G. M.
Jenkins, and G. C. Reinsel, 2013). The formulation is
as follows:
with λ is transformation parameter,
is
transformation data,
is Observation at time t
If the data are not stationary on mean, it is
necessary to do differencing. Backward shift operator
is very appropriate to describe the differencing
process (S. Makridakis, S. C. Wheelwright, and V. E.
McGee, 1999). The use of backshift is as follows :
(2)
BEST ICON 2018 - Built Environment, Science and Technology International Conference 2018
296