Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/75784
Title: Optimized integration algorithm model for stock price prediction with particle swarm optimization
Authors: Lin, Zhi
Keywords: DRNTU::Science::Mathematics::Discrete mathematics::Algorithms
Issue Date: 2018
Abstract: Stock occupies a very important position in the market economy. The individual can affect the operation of the company through the way of holdings. Stock can occupy a significant position in promoting the development of market economy as well. Hence, analysis and prediction on stock markets is not only helpful for the individual to make profit but also is benefit to the decision-makers on macroeconomic adjustment and maintain the development of the national economy. [1][2] With the developments of machine learning, many intelligent algorithms are springing up. According to the characteristics of the stock, it will produce great uncertainty in short-term investments, however, it meets the statistical regulation in the long-term trend. Therefore, in the case of infinite samples, predicting stock with machine learning algorithms is an important research direction. In this project, I proposed a new stock price prediction method. Building PSO-GBDT & PSO-RF model to solve the problems of stock prediction areas and obtain good performance from the combination of two algorithms. Selecting gradient boosting decision tree (GBDT) and random forest (RF) are the representative of integrated algorisms. Optimizing parameters of gradient boosting decision tree and random forest with particle swarm optimization. Building training stock price prediction model based on past performance of the Industrial Bank stock market from July 2017 to October 2017. All the data are provided by Straight flush and some of them are used to calculate the technical indicators. Last but not least, putting the test data into the debugged training model and get the predicted value. This project involves optimizing integrated algorisms and particle swarm optimization (PSO) with discussing the influence of each parameter on PSO. It is proofed that the combination of two algorithms has a good performance on stock prediction and PSO-RF model have achieved better results than PSO-GBDT model.
URI: http://hdl.handle.net/10356/75784
Schools: School of Mechanical and Aerospace Engineering 
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:MAE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
FYP Reprot Final.pdf
  Restricted Access
4.09 MBAdobe PDFView/Open

Page view(s)

280
Updated on Jun 22, 2024

Download(s) 50

22
Updated on Jun 22, 2024

Google ScholarTM

Check

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.