Penerapan Transformer pada Prediksi Indeks Harga Saham Gabungan (IHSG) Berdasarkan Data Makroekonomi, Indikator Teknis, dan Sentimen Berita Online

Styawan, Arif (2024) Penerapan Transformer pada Prediksi Indeks Harga Saham Gabungan (IHSG) Berdasarkan Data Makroekonomi, Indikator Teknis, dan Sentimen Berita Online. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002201035-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (3MB) | Request a copy

Abstract

Prediksi indeks harga saham seperti IHSG dapat membantu investor dalam mendapatkan keuntungan ketika berinvestasi di suatu pasar saham. Dengan mengetahui kondisi pasar, investor dapat mengetahui kinerja dari pasar saham tersebut. Akan tetapi, terdapat banyak faktor yang harus dipertimbangkan oleh investor dalam melakukan prediksi indeks harga saham. Oleh karena itu, tidak cukup hanya mempertimbangkan satu atau dua faktor saja. Banyaknya faktor yang berpengaruh, membuat investor harus melakukan analisis yang lebih mendalam. Di sisi lain, dengan adanya machine learning, dapat dimanfaatkan sebagai pendukung keputusan bagi investor ketika berinvestasi. Pada penelitian ini, dilakukan prediksi harga penutupan IHSG dengan menerapkan model Transformer menggunakan data-data yang berpengaruh terhadap fluktuasi IHSG. Data yang digunakan yaitu data makroekonomi, indikator teknis, dan sentimen berita online. Prediksi dilakukan pada harga penutupan (close price) IHSG pada hari berikutnya. Dataset yang digunakan dalam penelitian ini dimulai dari 3 Januari 2017 hingga 30 November 2023. Penelitian dilakukan menggunakan 4 variasi rasio dalam splitting data. Rasio pertama menggunakan perbandingan 70:15:15, rasio kedua menggunakan perbandingan 75:15:10, rasio ketiga menggunakan perbandingan 80:10:10, dan rasio keempat menggunakan perbandingan 85:10:5. Kinerja model dievaluasi menggunakan metrik MAE, RMSE, dan MAPE. Diperoleh hasil terbaik model dengan nilai MAE sebesar 40.760, RMSE sebesar 54.125, dan MAPE sebesar 0.591%.
==============================================================================================================================
Predicting stock market indices such as the IHSG can help investors gain profits when investing in a stock market. By understanding the market conditions, investors can assess the performance of the stock market. However, there are many factors that investors need to consider when predicting stock market indices. Therefore, it is not enough to consider just one or two factors. The multitude of influencing factors requires investors to conduct more in-depth analysis. On the other hand, with machine learning, investors can use it as a decision support tool when investing. In this study, the closing price prediction of the IHSG is performed by applying the Transformer model using data that affects IHSG fluctuations. The data used includes macroeconomic data, technical indicators, and online news sentiment. The prediction is carried out for the next day’s IHSG closing price. The dataset used in this study ranges from January 3, 2017, to November 30, 2023. The research was conducted using four variations of data splitting ratios. The first ratio used a 70:15:15 split, the second ratio used a 75:15:10 split, the third ratio used an 80:10:10 split, and the fourth ratio used an 85:10:5 split. The model’s performance was evaluated using the metrics MAE, RMSE, and MAPE. The best model results were obtained with an MAE of 40.760, an RMSE of 54.125, and a MAPE of 0.591%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Prediksi Indeks Harga Saham, Transformer, Time Series, IHSG, Stock Price Index Prediction
Subjects:	Q Science > QA Mathematics 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) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Arif Styawan
Date Deposited:	22 Aug 2024 02:49
Last Modified:	22 Aug 2024 02:49
URI:	http://repository.its.ac.id/id/eprint/114118

Actions (login required)

View Item