Peramalan Jumlah Kasus Tuberkulosis di Jawa Timur Menggunakan Hybrid Model ARIMA-NAR Neural Network

Melinda, Sarah Chairina (2019) Peramalan Jumlah Kasus Tuberkulosis di Jawa Timur Menggunakan Hybrid Model ARIMA-NAR Neural Network. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
05211540000015-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (7MB) | Request a copy

Abstract

Penyakit Tuberculosis (TB) merupakan penyakit infeksi pernafasan kronis yang disebabkan oleh Mycrobacterium tuberculosis dan menyebar melalui udara dari penderita. Penyebaran penyakit ini menjadi salah satu permasalahan kesehatan utama di dunia, termasuk pada daerah Jawa Timur, Indonesia, karena berbahaya dan menyebabkan kematian jika tidak ditangani dengan tepat waktu. Pada tahun 2016, jumlah penderita TB di Jawa Timur menempati peringkat ke-2 terbanyak di Indonesia, dengan jumlah 45.239. Tingginya angka ini menunjukkan perlunya Dinas Kesehatan Jawa Timur melakukan pencegahan dan pengendalian TB. Namun, kurangnya penelitian terkait prediksi jumlah kasus TB menyebabkan Pemerintah Daerah Jawa Timur tidak dapat menyusun prioritas waktu dan target implementasi pencegahan serta pengendalian TB secara tepat. Penelitian ini bertujuan untuk mengetahui kemungkinan tren dan prediksi jumlah kasus TB di Jawa Timur, mengetahui apakah Google Trends Search dapat menjadi data pendukung peramalan, serta mengetahui ketepatan Google Trends untuk menggantikan data kasus TB di Jawa Timur. Hasil penelitian ini berupa data peramalan dan akurasi peramalan time-series kasus tuberkulosis dari Dinas Kesehatan Jawa Timur dan Google Trends untuk tahun 2018 s.d. 2025 berdasarkan data triwulanan triwulan 1 tahun 2004 s.d. triwulan 3 2018. Peramalan dilakukan dengan menggunakan model ARIMA, ARIMAX dengan variabel bebas Google Trend, serta hybrid ARIMA-NAR NN dan ARIMAX NAR NN. Hasil evaluasi model menunjukkan bahwa penggunaan metode Hybrid memberikan hasil peramalan yang lebih baik dibandingkan dengan metode tunggal ARIMA dan ARIMAX, dikarenakan pola linear dan pola non-linear yang terdapat pada data akan cocok jika menggunakan metode Hybrid, yaitu ARIMA / ARIMAX untuk mengenali pola linear dan metode NAR untuk mengenali pola non-linear. Peningkatan kemampuan peramalan data Tuberkulosis Jawa Timur dari metode tunggal ARIMA ke metode Hybrid ARIMA-NAR ditunjukkan dengan rata-rata nilai penurunan MAPE sebesar 2,95% dan tunggal ARIMAX ke metode Hybrid ARIMAX-NAR dengan rata-rata nilai penurunan MAPE sebesar 2,43%. Metode ARIMAX-NAR memiliki kemampuan peramalan yang sedikit lebih kecil dibandingkan metode Hybrid ARIMA-NAR. Peningkatan kemampuan peramalan data Google Trend dari metode tunggal ARIMA ke metode ARIMA-NAR ditunjukkan dengan rata-rata penurunan MAPE sebesar 11,54%. Perbedaan akurasi peramalan yang tinggi antara model terbaik peramalan data Google Trend dengan data Tuberkulosis yaitu 14,36%, perbedaan range data yang signifikan dan kurangnya korelasi membuat Google Trend kurang dapat mewakili kasus Tuberkulosis di Jawa Timur. ================================================================================================ Tuberculosis (TB) is a chronic respiratory infection caused by Mycrobacterium tuberculosis and spreads through the air of the sufferer. The spread of this disease has become one of the major health problems in the world, including in the East Java region, Indonesia, because it is dangerous and causes death if not handled in a timely manner. In 2016, the number of TB patients in East Java was ranked as the second largest in Indonesia, with a total of 45,239. This high number indicates the need for the East Java Health Office to prevent and control TB. However, the lack of research related to the prediction of the number of TB cases led to the East Java Regional Government not being able to prioritizing and targetting the implementation of TB prevention and control appropriately. This study aims to determine the possible trends and predictions of the number of TB cases in East Java, find out whether Google Trends Search can be a supporting data for forecasting, as well as knowing the accuracy of Google Trends to replace TB case data in East Java. The results of this study are forecasting data and accuracy of tuberculosis case time-series forecasting from the East Java Health Office and Google Trends for 2018.d. 2025 based on quarterly data from 1st quarter 2004 untill 3rd quarter 2018. Forecasting is done using the ARIMA model, ARIMAX with Google Trend independent variables, and hybrid ARIMA-NAR NN and ARIMAX NAR NN. The model evaluation results show that the use of the Hybrid method gives better forecasting results than the single ARIMA and ARIMAX methods, because linear patterns and non-linear patterns contained in the data would be suitable if using the Hybrid method, namely ARIMA / ARIMAX to recognize linear patterns and the NAR method for recognizing non-linear patterns. The increase in the forecasting ability of East Java Tuberculosis data from the ARIMA single method to the Hybrid ARIMA-NAR method is indicated by the average value of MAPE reduction of 2.95% and single ARIMAX to the Hybrid ARIMAX-NAR method with an average MAPE decrease of 2.43 %. The ARIMAX-NAR method has a forecasting capability that is slightly smaller than the Hybrid ARIMA-NAR method. The increase in Google Trend data forecasting capability from the ARIMA single method to the ARIMA-NAR method is shown by the average MAPE reduction of 11.54%. The high difference in forecasting accuracy between the best models of Google Trend data forecasting and Tuberculosis data is 14.36%, significant differences in data range and lack of correlation make Google Trend less representative of Tuberculosis cases in East Java.

Item Type: Thesis (Undergraduate)
Additional Information: RSSI 519.535 Mel p-1 2019
Uncontrolled Keywords: Tuberkulosis, Peramalan, Google Trends, ARIMA, ARIMAX, NAR, Neural Network, Jawa Timur
Subjects: Q Science > Q Science (General)
Q Science > Q Science (General) > Q325 GMDH algorithms.
Q Science > Q Science (General) > Q325.78 Back propagation
Q Science > Q Science (General) > Q337.3 Swarm intelligence
Q Science > Q Science (General) > Q337.5 Pattern recognition systems
Q Science > QA Mathematics > QA280 Box-Jenkins forecasting
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User: Sarah Chairina Melinda
Date Deposited: 20 Sep 2021 02:55
Last Modified: 20 Sep 2021 02:55
URI: https://repository.its.ac.id/id/eprint/60509

Actions (login required)

View Item View Item