Analisis Performansi Optimasi Long Short Term Memory (LSTM) Menggunakan Hasil Analisa Sentimen

Amanullah, Nurkholis (2024) Analisis Performansi Optimasi Long Short Term Memory (LSTM) Menggunakan Hasil Analisa Sentimen. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 20081010046.-cover.pdf Download (1MB)
	Text (Bab 1) 20081010046.-bab1.pdf Download (256kB)
	Text (Bab 2) 20081010046.-bab 2.pdf Restricted to Repository staff only until 4 June 2028. Download (679kB)
	Text (Bab 3) 20081010046.-bab 3.pdf Restricted to Repository staff only until 4 June 2028. Download (446kB)
	Text (Bab 4) 20081010046.-bab 4.pdf Restricted to Repository staff only until 4 June 2028. Download (1MB)
	Text (Bab 5) 20081010046.-bab 5.pdf Download (97kB)
	Text (Dapus) 20081010046.-daftarpustaka.pdf Download (243kB)
	Text (Lampiran) 20081010046.-lampiran.pdf Restricted to Repository staff only until 4 June 2028. Download (2MB)

Abstract

The stock market is a complex and uncertain arena yet a highly desirable financial instrument. financial instrument that is in high demand. Trading stocks, binaries, gold, and bitcoin are growing in popularity, but carry the risk of significant price fluctuations influenced by economic and political factors. influenced by economic and political factors. Growing interest in capital market investments, especially stocks, drives the need for additional information for investors. for investors. Social media, particularly Twitter, has become a place to share views and opinions about the company. and opinions related to the company. Social media sentiment analysis can provide additional insights to evaluate potential future stock price movements, preventing unwanted speculation. future, preventing unwanted speculation. This research uses block diagrams to analyze the stock of Tesla stock using the Long Short-Term Memory (LSTM) method and social media sentiment analysis from the Twitter platform. social media sentiment analysis from the Twitter platform. Tesla stock price data is obtained from Kaggle, while Twitter sentiment data is processed through a pre-processing stage. The results of the data analysis show the variation of the results of the various variables in a scheme. Evaluation of 54 LSTM models with split parameters of the amount of data, division of train and test data, number of hidden layers, neurons and learning rate, revealed significant differences. revealed significant differences, which confirmed the importance of adjustment of variable values to achieve an optimal model. Data integration LSTM and sentiment data expands the understanding of the factors that influencing Tesla's stock price. The results show a significant improvement in the significant improvement in the accuracy of the LSTM model after integration, The optimal configuration of the LSTM configuration found includes a learning rate of 0.01, a data split of 90%:10%, and the use of 2 layers with 50 and 100 neurons respectively, resulting in an MSE of 63.33, RMSE of 7.95, and MAPE of 2.26%. The test results get a range of MAPE values <10%, so this model is an excellent forecasting model. including a very good forecasting model. Social media sentiment integration is also proven to improve model accuracy, as evidenced by an increase in MSE by 62.88, RMSE of 7.92 with a MAPE of almost 0.02%. Although this research is limited to one stock, suggestions for the development of a wider dataset are development of a wider dataset, universal model development, and benefits for stock market decision makers. for stock market decision makers are important points for future research. future research. Keywords: LSTM, forecasting, time series, sentiment analysis, integration, Tesla, Twitter

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Diyasa, I Gede Susrama Mas	NIDN0019067008	igsusrama.if@upnjatim.ac.id
Thesis advisor	Rizki, Agung Mustika	NIDN0025079302	agung.mustika.if@upnjatim.ac.id

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Nurkholis Amanullah

Date Deposited:

04 Jun 2024 02:01

Last Modified:

04 Jun 2024 02:01

URI:

https://repository.upnjatim.ac.id/id/eprint/23939