Peramalan Harga Saham PT Indofood Sukses Makmur Tbk. (INDF) Menggunakan Metode Hybrid Informer, Convolutional Neural Network (CNN) Dan Long Short-term Memory (LSTM)

Putri, Sintiarani Febyan (2026) Peramalan Harga Saham PT Indofood Sukses Makmur Tbk. (INDF) Menggunakan Metode Hybrid Informer, Convolutional Neural Network (CNN) Dan Long Short-term Memory (LSTM). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5026221044_Undergraduate_Thesis.pdf Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Peramalan harga saham merupakan aspek penting dalam pengambilan keputusan investasi berbasis data-driven decision making, terutama pada pasar modal yang bersifat volatil dan non-linear. Pendekatan prediksi konvensional masih memiliki keterbatasan dalam memahami pola pergerakan harga saham yang kompleks dan dinamis. Oleh karena itu, tugas akhir ini bertujuan untuk membangun dan mengevaluasi model peramalan harga saham harian PT Indofood Sukses Makmur Tbk. (INDF) dengan mengusulkan pendekatan hybrid sequential Informer–CNN–LSTM. Informer digunakan untuk memodelkan long-term dependencies, CNN mengekstraksi local temporal patterns, sedangkan LSTM mempelajari dinamika short-term pada data harga saham. Penelitian ini menggunakan pendekatan univariat dengan variabel closing price berdasarkan data harga saham harian INDF periode Oktober 2015 hingga Oktober 2025 yang diperoleh dari Investing.com. Tahapan penelitian meliputi preprocessing, perancangan model, proses training, dan evaluasi menggunakan metrik MAE, RMSE, MAPE, serta koefisien determinasi (R²). Hasil pengujian menunjukkan bahwa model CNN memberikan performa terbaik dengan nilai MAPE 1,0486%, RMSE 107,81, MAE 75,20, dan R² 0,9782. Sementara itu, model Hybrid Informer–CNN–LSTM memperoleh performa terbaik dengan nilai MAPE 1,2824%, RMSE 125,89, MAE 91,98, dan R² 0,9703. Hasil penelitian menunjukkan bahwa penambahan arsitektur hybrid tidak selalu menghasilkan akurasi yang lebih tinggi dibandingkan model tunggal. Meskipun demikian, model Hybrid Informer–CNN–LSTM tetap mampu menghasilkan prediksi yang akurat sehingga dapat menjadi alternatif dalam pengembangan sistem pendukung keputusan investasi berbasis business intelligence.
=======================================================================================================================================
Stock price forecasting is an important aspect of investment decision-making based on data-driven decision making, particularly in the highly volatile and non-linear capital market. Conventional forecasting approaches still face limitations in capturing the complex and dynamic patterns of stock price movements. Therefore, this undergraduate thesis aims to develop and evaluate a daily stock price forecasting model for PT Indofood Sukses Makmur Tbk. (INDF) using a hybrid sequential Informer–CNN–LSTM approach. The Informer is used to model long-term dependencies, CNN is applied to extract local temporal patterns, while LSTM captures short-term dynamics in stock price data. This study adopts a univariate approach using the closing price as the main variable, based on daily INDF stock price data from October 2015 to October 2025 obtained from Investing.com. The research process includes data preprocessing, model design, training, and evaluation using MAE, RMSE, MAPE, and the coefficient of determination (R²). Experimental results show that the CNN model achieves the best performance with a MAPE of 1.0486%, RMSE of 107.81, MAE of 75.20, and R² of 0.9782. Meanwhile, the Hybrid Informer–CNN–LSTM model achieves its best performance with a MAPE of 1.2824%, RMSE of 125.89, MAE of 91.98, and R² of 0.9703. The results indicate that adding a hybrid architecture does not always lead to better accuracy compared to a single model. However, the Hybrid Informer–CNN–LSTM model is still capable of producing accurate predictions and can serve as an alternative approach for developing investment decision support systems based on business intelligence.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Hybrid Deep Learning, Informer–CNN–LSTM, Stock Price Forecasting, Time Series, Hybrid Deep Learning, Informer–CNN–LSTM, Peramalan Harga Saham, Time Series
Subjects:	Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Sintiarani Febyan Putri
Date Deposited:	30 Jul 2026 03:27
Last Modified:	30 Jul 2026 03:27
URI:	http://repository.its.ac.id/id/eprint/140217

Actions (login required)

View Item