Prediksi Harga Saham PT. Bank Rakyat Indonesia (Persero) Tbk Berdasarkan Analisis Sentimen Dan IHSG Menggunakan Metode Long-Short Term Memory

Martiza, Jasmine Rayra (2026) Prediksi Harga Saham PT. Bank Rakyat Indonesia (Persero) Tbk Berdasarkan Analisis Sentimen Dan IHSG Menggunakan Metode Long-Short Term Memory. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2043221143-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Harga saham merupakan representasi nilai suatu perusahaan yang mencerminkan kinerja perusahaan di masa mendatang. Pergerakan harga saham dipengaruhi oleh berbagai faktor, baik yang berasal dari kondisi internal perusahaan maupun faktor eksternal seperti sentimen pasar dan kondisi ekonomi secara umum. Fluktuasi harga saham mendorong perlunya prediksi yang akurat sebagai dasar pengambilan keputusan investasi. PT Bank Rakyat Indonesia (Persero) Tbk merupakan salah satu emiten perbankan dengan tingkat likuiditas tinggi di pasar modal Indonesial. Meskipun demikian, harga saham BBRI menunjukkan tren yang fluktuatif pada periode 2022 -2025. Oleh karena itu, penelitian ini bertujuan memprediksi harga saham BBRI dengan mempertimbangkan sentimen pasar dan Indeks Harga Saham Gabungan (IHSG). Analisis sentimen dilakukan menggunakan model IndoRoBERTa, kemudian hasil sentimen, IHSG dan harga saham diprediksi menggunakan metode Long Short-Term Memory (LSTM). Hasil penelitian menunjukkan bahwa sentimen berita terhadap saham BBRI didominasi oleh sentimen positif sebanyak 3.720 berita, yang mengindikasikan bahwa pemberitaan selama periode penelitian cenderung bernada positif. Berdasarkan perbandingan empat skenario pemodelan yang digunakan dalam penelitian, Model B yang memprediksi harga saham berdasarkan harga saham periode sebelumnya serta variabel IHSG pada periode saat ini dan periode sebelumnya terpilih sebagai model terbaik dengan nilai RMSE sebesar 73,272 dan MAPE sebesar 1,527%, hasil tersebut menunjukkan bahwa penambahan variabel IHSG mampu meningkatkan akurasi prediksi pada harga saham BBRI. Selanjutnya hasil peramalan menggunakan model terbaik menghasilkan nilai MAPE sebesar 13,4% yang menunjukkan tingkat akurasi prediksi yang baik. Meskipun demikian, model masih memiliki keterbatasan dalam mengikuti pola pergerakan harga saham secara optimal, khususnya dalam menangkap tren penurunan yang terjadi selama periode peramalan.
=================================================================================================================================
Stock price represents a company's value and reflects its future performance. Stock price movements are influenced by various factors, including internal company conditions and external factors such as market sentiment and overall economic conditions. The fluctuating nature of stock prices highlights the need for accurate forecasting to support investment decision-making. PT Bank Rakyat Indonesia (Persero) Tbk is one of the most liquid banking stocks in the Indonesian capital market. Nevertheless, BBRI's stock price exhibited considerable fluctuations during the 2022–2025 period. Therefore, this study aims to forecast BBRI's stock price by incorporating market sentiment and the Indonesia Composite Stock Price Index (IHSG). Sentiment analysis was performed using the IndoRoBERTa model, while stock price forecasting was conducted using the Long Short-Term Memory (LSTM) method based on sentiment scores, the IHSG, and historical stock prices. The results indicate that news sentiment related to BBRI was dominated by positive sentiment, accounting for 3.720 news articles, suggesting that the overall news coverage during the study period was predominantly positive. Among the four modeling scenarios evaluated, Model B, which utilizes the previous stock price together with the current and previous IHSG values, achieved the best performance with an RMSE of 73,272 and a MAPE of 1,527%. These findings indicate that incorporating the IHSG variable improves the forecasting accuracy of BBRI's stock price. Furthermore, the forecasting results obtained using the best-performing model yielded a MAPE of 13,4%, indicating good predictive accuracy. However, the model still has limitations in accurately capturing stock price movements, particularly in following the downward trend observed during the forecasting period.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	BBRI, IHSG, IndoRoBERTa, Long Short Term Memory, Sentimen Berita BBRI, IDX Composite, IndoRoBERTa, Long Short-Term Memory, News Sentiment
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics Q Science > QA Mathematics > QA336 Artificial Intelligence
Divisions:	Faculty of Vocational > 49501-Business Statistics
Depositing User:	Jasmine Rayra Martiza
Date Deposited:	13 Jul 2026 07:02
Last Modified:	13 Jul 2026 07:02
URI:	http://repository.its.ac.id/id/eprint/134795

Actions (login required)

View Item