MODEL ARIMA (p, d, q) PADA DATA KEMATIAN IBU HAMIL (STUDI KASUS DI RSUD ULIN BANJARMASIN)

Abstract

Time series analysis is a sequence of quantitative observation at a certain time
interval, such as daily, monthly, yearly, etc. This analysis has been used for
identifying the characteristic, pattern, and model of observed data in the specified
past period that leads to predict the future observation values. Applications of
time series analysis include all aspects such as, meteorology, economy, and
medical area. The deaths prevalence of pregnancy women at delivering process is
still very high. This leads to a significant medical problem in Indonesia. In 2003,
this case has been reported 20 women, including 3 cases due to blooding, 3 cases
due to eclampsy, and 14 cases due to other factors. This research is conducted to
give an appropriate model to predict the deaths of pregnancy women in the future.
The method of this research is using purposive sampling technique, which is
gathering 72 secondary data of monthly deaths pregnancy women from
2003 - 2008 in the midwife room of Ulin Hospital, Banjarmasin. The experiment
procedures are identifying the appropriate ARIMA model to give the trend
monthly deaths description of pregnancy women, and evaluating the fitted ARIMA
model to forecast monthly deaths of pregnancy women in the future 10 years
based on the given historical data.
Based on the conducted analysis, it is found that monthly deaths of pregnancy
women between 2003 and 2008 in Ulin Hospital, Banjarmasin has experienced
a slightly increase. The data are discrete with monthly expected value equal to
0.625, where the smallest and biggest deaths are 0 and 3 people, respectively.
Beside this, it has been investigated that ARIMA (0, 0, 0) or ARMA (0, 0) is the
best ARIMA model for this case because it results the minimum value of AICC.