Analisis Pergerakan Harga Saham PT Unilever Indonesia Tbk Berdasarkan Sentimen Publik Menggunakan Machine Learning

Firdaus, Dinda Apriliyani (2025) Analisis Pergerakan Harga Saham PT Unilever Indonesia Tbk Berdasarkan Sentimen Publik Menggunakan Machine Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2043211056-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (1MB) | Request a copy

Abstract

Penelitian ini bertujuan untuk menganalisis pengaruh sentimen publik di media sosial terhadap pergerakan harga saham PT Unilever Indonesia Tbk dengan pendekatan machine learning dan time series forecasting. Data dikumpulkan melalui crawling media sosial X (Twitter) selama periode 2023–2024 dan menghasilkan 6.903 tweet yang berkaitan dengan isu boikot dan persepsi publik terhadap Unilever. Tweet tersebut diproses menggunakan metode leksikon untuk pelabelan sentimen, lalu dikonversi menjadi variabel numerik harian berupa variabel dummy (D) dan diintegrasikan ke dalam model prediksi. Penelitian ini membandingkan dua model, yaitu Support Vector Regression (SVR) dan ARIMAX. Model SVR dengan input Yt-1 (harga saham hari sebelumnya) dan D menunjukkan bahwa sentimen publik memiliki pengaruh terhadap pergerakan harga saham, dengan nilai MAPE sebesar 5.50%. Sementara itu, model ARIMAX dengan input Yt-1 dan Yt-2 memberikan hasil prediksi paling akurat dengan MAPE sebesar 2.35%. Visualisasi prediksi melalui aplikasi RShiny menunjukkan bahwa kedua model mampu menghasilkan proyeksi harga saham yang mendekati nilai aktual, meskipun pendekatan ARIMAX lebih unggul dalam akurasi, sedangkan SVR mampu mengakomodasi faktor eksternal berupa sentimen publik.
===================================================================================================================================
This study aims to analyze the influence of public sentiment on social media toward the stock price movement of PT Unilever Indonesia Tbk using machine learning and time series forecasting approaches. Data were collected through web crawling from X (Twitter) during the 2023–2024 period, resulting in 6,903 tweets related to boycott issues and public perceptions of Unilever. The tweets were processed using a lexicon-based sentiment analysis method and converted into daily numeric variables in the form of dummy sentiment variables (D), which were then integrated into predictive models. This research compares two models: Support Vector Regression (SVR) and ARIMAX. The SVR model, using Yt-1 (previous day’s stock price) and D as inputs, indicates that public sentiment influences stock price movement, achieving a MAPE of 5.50%. In contrast, the ARIMAX model, using Yt-1 and Yt-2 as inputs, shows the highest prediction accuracy with a MAPE of 2.35%. Visualization using an RShiny application demonstrates that both models produce stock price predictions close to the actual values, with ARIMAX showing superior accuracy, while SVR excels in integrating external factors such as public sentiment.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	ARIMAX, Boikot, Crawling, Harga Saham, Sentimen, Media Sosial, SVR, Unilever, Boycott, Stock Price.
Subjects:	Q Science Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Vocational > 49501-Business Statistics
Depositing User:	Dinda Apriliyani Firdaus
Date Deposited:	18 Nov 2025 05:59
Last Modified:	18 Nov 2025 05:59
URI:	http://repository.its.ac.id/id/eprint/128786

Actions (login required)

View Item