TLKM STOCK PRICE PREDICTION USING A HYBRID LINEAR REGRESSION AND DEEP NEURAL NETWORK MODEL

Haqiqi, Muhammad Sulthon (2026) TLKM STOCK PRICE PREDICTION USING A HYBRID LINEAR REGRESSION AND DEEP NEURAL NETWORK MODEL. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 22081010208.-cover.pdf Download (821kB)
	Text (Bab 1) 22081010208.-bab1.pdf Download (161kB)
	Text (Bab 2) 22081010208.-bab2.pdf Restricted to Repository staff only until 2 September 2029. Download (667kB)
	Text (Bab 3) 22081010208.-bab3.pdf Restricted to Repository staff only until 2 September 2029. Download (318kB)
	Text (Bab 4) 22081010208.-bab4.pdf Restricted to Repository staff only until 2 September 2029. Download (815kB)
	Text (Bab 5) 22081010208.-bab5.pdf Download (107kB)
	Text (Daftar Pustaka) 22081010208.-daftarpustaka.pdf Download (159kB)

Abstract

Stock price movements are dynamic, highly volatile, and influenced by various economic factors, making stock price prediction a challenging task. Conventional prediction methods often have limitations in capturing both linear and non-linear relationships within stock market data, leading to the need for more effective predictive approaches. This study proposes a hybrid model that combines Linear Regression and an Ensemble Deep Neural Network (DNN) to predict the stock price of PT Telkom Indonesia Tbk (TLKM). Linear Regression is employed to capture the primary linear trends in the data, while the Ensemble DNN is utilized to learn non-linear residual patterns and improve prediction accuracy. The model is further enhanced by incorporating macroeconomic variables, including the USD/IDR exchange rate, inflation rate, and BI Rate. The research process consists of data collection, data preprocessing, model training, and the implementation of a web based system using Streamlit. To ensure the reliability of the model when dealing with time-series data, the evaluation process applies Time Series Cross-Validation, which partitions the data chronologically to prevent data leakage and simulate different market conditions. The results indicate that the proposed hybrid model achieves strong predictive performance, with an average Mean Absolute Percentage Error (MAPE) of 8.98%, a Mean Absolute Error (MAE) of IDR 262.61, and a Root Mean Squared Error (RMSE) of IDR 322.00. These findings demonstrate that the combination of Linear Regression and Ensemble Deep Neural Network effectively improves the accuracy of TLKM stock price prediction while maintaining relatively stable performance across different validation scenarios. Keywords: TLKM Stock, Linear Regression, Deep Neural Network, Hybrid Model, Stock Prediction.

Item Type:

Thesis (Undergraduate)

Contributors: