Prediksi Harga Saham Yang Bersifat Siklikal Di Indonesia Menggunakan Metode LSTM dan SVM

Gabriel Adisurya Harsono, Alexander Setiawan, Hans Juwiantho

Abstract


Stock Market prediction is not an easy task, eventhou using technical analysis. Technical analysis is a technique that can predict stock market price. But using this technique is not an easy task, hence there is a change of mistake in analyzing the stoks that create losses for investors and traders. This is where machine learning and deep learning are able to predict a stock that have cyclical behavior. Cyclical stocks have patterns that other stocks do not have, where machine learning and deep learning model are able to catch and learn the pattern on cyclical stocks and predict the stock prices. The prediction are used to give recommandation for the users when to but or when to sell stocks, but it also able to give recommandation and show stock prices a year in the future. In this Undergraduatae Thesis will be used five methode and two type of data, this include LSTM, Bi-LSTM, GRU, SARIMAX, and SVR, the data contains multivariate and univariate type of data. Multivariate data will contain the stock prices, Jakarta composite index, and the support data close prices. The result of prediction will be compared with the existing data using RMSE then compare the result with all existing model. In the proses of the undergraduate thesis, every parameters in the model will be searched using GridsearchCV and the best parameter for each models. From this research found 1 best metode for multivariate prediction and one for univariate. for multivariate the best parameter for the prediction is LSTM where the RMSE value is 63.67 training and 74.82 for testing. For univariate prediction the best metode is SVR where the RMSE value is 58.04 for training and 75.29 for testing.

Keywords


Stock Market; RNN; Deep Learning; Machine Learning

Full Text:

PDF

References


Althelaya, Khaled A., El-Sayed M. El-Alfy, and Salahadin

Mohammed. "Stock market forecast using multivariate

analysis with bidirectional and stacked (LSTM,

GRU)." 2018 21st Saudi Computer Society National

Computer Conference (NCC). IEEE, 2018. DOI:

1109/NCG.2018.8593076

Fischer, T., & Krauss, C. (2017). Deep learning with long

short-term memory. Discussion Papers.

Hochreiter, S., & Schmidhuber, J. (1997). Long Short-Term

Memory. DOI: 10.1162/neco.1997.9.8.1735

Hakob, G. (2016). A Stock Market Prediction Method

Based on Support Vector Machines.

Karmiani, D., Kazi, R., Nambisan, A., Shah, A., & Kamble,

V. (2019). Comparison of Predictive Algorithms:

Backpropagation, SVM, LSTM and Kalman Filter for Stock

Market. DOI: 10.1109/AICAI.2019.8701258

Lasek, A., Cercone, N., & Saunders, J. (2016). SMART

RESTAURANTS. SOURVEY ON CUSTOMER DEMAND

AND SALES FORCASTING. DOI: 10.1016/B978-0-12-

-5.00017-1

Nikou, M., Mansourfar, G., & Bagherzadeh, J. (2019).

Stock price prediction using DEEP learning algorithm and

its. DOI: 10.1002/isaf.1459

Roy, K., Kar, S., Narayan, R., & Das. (2015). Chapter 6 -

Selected Statistical Methods in QSAR. Understanding the

Basics of QSAR for Applications in Pharmaceutical

Sciences and Risk Assessment, 191-229.

Siami-Namini, S., Tavakoli, N., & Namin, A. S. (2018). A

Comparison of ARIMA and LSTM in Forecasting Time

Series. DOI: 10.1109/ICMLA.2018.00227

Viadinugroho, R. A., & Rosadi, D. (2021). Long ShortTerm Memory Neural Network Model for Time. Journal of

Physics: Conference Series. ISSN: 2613-9189


Refbacks

  • There are currently no refbacks.


Jurnal telah terindeks oleh :