Patricia, Jennifer (2021) Peramalan Laju Produk Domestik Bruto Indonesia dengan Data Google Trends Menggunakan Metode Neural Network dan eXtreme Gradient Boosting. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211740000037-Undergraduate_Thesis.pdf - Accepted Version Download (3MB) | Preview |
Abstract
Aktivitas ekonomi yang tidak menentu dikarenakan ketergantungannya dengan kondisi pandemi COVID-19 menjadikan indikator makro ekonomi seperti laju Produk Domestik Bruto memiliki peran yang sangat penting dalam membantu para pembuat kebijakan maupun pebisnis untuk mengerti kondisi perekonomian negara. Tujuan dari penelitian ini ialah meramalkan laju Produk Domestik Bruto dalam periode bulanan dengan lag data google trends, indikator ekonomi, dan official statistics. Metode yang digunakan pada penelitian ini adalah metode machine learning Neural Network (NN) dan Extreme Gradient Boosting (XGBoost), dimana kedua metode tersebut merepresentasikan pemodelan pola data dense dan sparse secara berurutan. Hasil eksplorasi data menunjukkan nilai laju PDB Indonesia mengalami penurunan secara drastis dari saat timbulnya kasus positif COVID-19 pertama di Indonesia sehingga model ARIMA yang digunakan sebagai metode pembanding dari kedua metode machine learning adalah model ARIMA Intervensi. Dari analisis yang dilakukan diketahui bahwa model NN dan XGBoost memberikan pilihan variabel input yang berbeda untuk mendapatkan hasil ramalan terbaik berdasarkan nilai variable importance. Metode XGBoost dengan tuning hyperparameter menjadi metode terbaik dalam meramalkan nilai laju PDB dengan nilai RMSE dan SMAPE pada data testing secara berturut sebesar 0,309 dan 33,461%.
=====================================================================================================
Uncertainty in economic activity due to its dependency with COVID-19 pandemic condition makes macroecomic indicators, such as Gross Domestic Product (GDP) growth, play a huge role in helping policy makers and business owners to understand the economy of a country. The purpose of this study is to forecast monthly Indonesian GDP growth using the lag of google trends, economic indicator, and official statistics data as the input variables. Machine learning methods that are used to forecast are Neural Network (NN) and eXtreme Gradient Boosting (XGBoost), where both methods represent dense and sparse modelling algorithm respectively. The result of exploratory data analysis showed that Indonesian GDP growth decreased significantly when the first positive case of COVID-19 appeared in Indonesia, therefore ARIMA model that will be used as benchmark for the other models is ARIMA with intervention analysis. Based on the analysis that has been done, the choice of input variables for NN and XGBoost model are different due to the variable importance values for each model. XGBoost model with hyperparameter tuning is the best model to forecast GDP growh with accuracy of testing data 0,309 for RMSE and 33,461% for SMAPE
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | Big data, google trends, NN, forecasting, PDB, XGBoost, GDP |
| Subjects: | Q Science > QA Mathematics Q Science > QA Mathematics > QA280 Box-Jenkins forecasting Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Jennifer Patricia |
| Date Deposited: | 31 Aug 2021 06:04 |
| Last Modified: | 10 Jun 2024 00:53 |
| URI: | http://repository.its.ac.id/id/eprint/90922 |
Actions (login required)
 |
View Item |