SVM Analysis for 5G Stock Movement Prediction

Author: J. Francisco Salazar G

Project Overview

This repository hosts a comprehensive analysis using Support Vector Machine (SVM) to predict the directional movement of stocks related to the emerging 5G technology. It covers the period from January 2019 to October 2020, focusing on the historical price data of two major 5G-related stocks.

Introduction

This project investigates whether historical price data can help us predict the price direction of stocks within the 5G industry. It treats stock price forecasting as a binary classification problem and utilizes SVM as the learning technique to forecast future price movements.

Features of the Repository

• Jupyter notebooks with in-depth mathematical explanations of SVM, hyperplanes, and margin equations.
• Analysis of the hard margin versus soft margin classification in SVM.
• Evaluation of the model's performance using ROC, confusion matrix, and other metrics.
• A detailed exploratory data analysis section with visualizations of the stock price series.
• Scripts for data preprocessing, including feature scaling and selection.
• A section on feature selection to enhance model accuracy.

Usage

To run the notebooks and scripts in this repository:

git clone https://github.com/FranQuant/Predicting-Asset-Move-Direction-using-Support-Vector-Machines.git

Contributing

Feedback, issues, and pull requests are welcome to improve the methodologies or explore new strategies.

License

This project is licensed under the MIT License - see the LICENSE.md file for details.