THE, MICHAEL (2024) COMPARISON K-NN REGRESSION AND SVR MSE IN 5 MOST ACTIVE STOCK BASED ON HISTORICAL DATA. S1 thesis, UNIVERSITAS KATOLIK SOEGIJAPRANATA.

Preview	Text 21.K1.0014_MICHEL THE_COVER.pdf Download (307kB) | Preview
	Text 21.K1.0014_MICHEL THE_BAB I.pdf Restricted to Registered users only Download (172kB)
	Text 21.K1.0014_MICHEL THE_BAB II.pdf Restricted to Registered users only Download (153kB)
	Text 21.K1.0014_MICHEL THE_BAB III.pdf Restricted to Registered users only Download (419kB)
	Text 21.K1.0014_MICHEL THE_BAB IV.pdf Restricted to Registered users only Download (835kB)
	Text 21.K1.0014_MICHEL THE_BAB V.pdf Restricted to Registered users only Download (142kB)
Preview	Text 21.K1.0014_MICHEL THE_DAPUS.pdf Download (140kB) | Preview
	Text 21.K1.0014_MICHEL THE_LAMP.pdf Restricted to Registered users only Download (2MB)

Abstract

This project aims to address the challenge of accurate stock price prediction by comparing the error prediction of two popular machine learning algorithms, K-Nearest Neighbors Regressor (KNN-R) and Support Vector Regressor (SVR) based on historical stock data. Based on previous research about stock prediction, the study evaluates the error in predicting future stock prices based on features such as historical price and volume. The SVR is evaluated for its ability to fit a regression model that minimizes prediction errors. In terms of accuracy, KNN Regression is hypothesized to outperform SVR. The models are trained and rested across multiple stock datasets, and results indicate that SVR consistently achieves superior predictive performance compared to KNN Regression. These findings highlight the importance of selecting the right algorithm for stock price prediction.

Actions (login required)

View Item