BUMI Stock Price Prediction Using Long Short Term Memory (LSTM) with Three Hyperparameter Tuning Regression

Hartono Hartono

doi:10.35877/454RI.qems1118

Authors

Hartono Universitas Widya Dharma Pontianak, Indonesia

DOI:

https://doi.org/10.35877/454RI.qems1118

Keywords:

LSTM, BUMI, Shares, Forecasting

Abstract

Shares is one of investment with moderate to high level of risk profile and return rate, and is a legal ownership proof of a company (Ltd). To obtain an optimum return with possible low risk, price of shares can be predicted or analyzed via fundamental or technical analysis. One of the most popular technical analysis is by employing computer algorithms as a device to help making decision. Long short term memory (LSTM), as a deep improvement of recurrent neural networks (RNN), is a commonly used method to predict shares price. BUMI, as one of shares in the energy sector, is the most concerned shares due to the geopolitical conflict in Europe. This research aims to predict BUMI shares price using LSTM by considering three parameters, i.e. number of hidden layers, epoch, and learning rate. Linear regression is used to obtained optimum values of each parameter whichs further combined to yield the best performance of the algorithm. Based on the analysis, the combination of minimum and optimum parameters results in the best performance by RMSE value.

References

[1] S. Hermuningsih, “Pengantar Pasar Modal Indonesia,” Yogyak. UPP STIM YKPN, 2012.
[2] R. B. Purnama, M. Malyadi, and N. R. Novawati, “Perancangan prediksi untuk menentukan indeks harga saham menggunakan jaringan syaraf tiruan,” Kinet. Game Technol. Inf. Syst. Comput. Netw. Comput. Electron. Control, pp. 125–130, 2017.
[3] Khalis Sofi, Aswan Supriyadi Sunge, Sasmitoh Rahmad Riady, and Antika Zahrotul Kamalia, “Perbandingan algoritma linear regression, LSTM, dan GRU dalam memprediksi harga saham dengan model time series,” SEMINASTIKA, vol. 3, no. 1, pp. 39–46, Nov. 2021, doi: 10.47002/seminastika.v3i1.275.
[4] R. Julian and M. R. Pribadi, “Peramalan harga saham pertambangan pada bursa efek indonesia (BEI) menggunakan long short term memory (LSTM),” JATISI J. Tek. Inform. Dan Sist. Inf., vol. 8, no. 3, pp. 1570–1580, Sep. 2021, doi: 10.35957/jatisi.v8i3.1159.
[5] A. Agusta, I. Ernawati, and A. Muliawati, “Prediksi pergerakan harga saham pada sektor farmasi menggunakan algoritma long short-term memory,” Inform. J. Ilmu Komput., vol. 17, no. 2, p. 164, Dec. 2021, doi: 10.52958/iftk.v17i2.3651.
[6] A. Arfan and L. Etp, “Perbandingan algoritma long short-term memory dengan SVR pada prediksi harga saham di Indonesia,” PETIR, vol. 13, no. 1, pp. 33–43, Mar. 2020, doi: 10.33322/petir.v13i1.858.
[7] L. E. Siahaan, “Prediksi Indeks Harga Saham dengan Metode Gabungan Support Vector Regression dan Jaringan Syaraf Tiruan,” Indones. J. Comput. Indo-JC, vol. 2, no. 1, pp. 21–30, 2017.
[8] Y. Baek and H. Y. Kim, “ModAugNet: A new forecasting framework for stock market index value with an overfitting prevention LSTM module and a prediction LSTM module,” Expert Syst. Appl., vol. 113, pp. 457–480, 2018.
[9] S. Mutasa, S. Sun, and R. Ha, “Understanding artificial intelligence based radiology studies: What is overfitting?,” Clin. Imaging, vol. 65, pp. 96–99, 2020.
[10] J. Schmidhuber, “Discovering neural nets with low Kolmogorov complexity and high generalization capability,” Neural Netw., vol. 10, no. 5, pp. 857–873, 1997.
[11] R. Faurina, B. Winduratna, and P. Nugroho, “Predicting Stock Movement Using Unidirectional LSTM and Feature Reduction: The Case of An Indonesia Stock,” Kuta, Bali, Indonesia, Nov. 2014.