This project focuses on predicting stock returns using time series analysis and machine learning. The goal is to build a regression model that can forecast future stock returns based on historical data. Visit the page.
Two main approaches were taken in this project:
- ARIMA (AutoRegressive Integrated Moving Average): Utilized for modeling and forecasting the time series data of stock returns.
- The ARIMA model was used to analyze the temporal structure of the stock returns and to generate short-term forecasts.
- LightGBM (Light Gradient Boosting Machine): Employed for its efficiency and accuracy in handling large datasets and its importance in feature selection.
- Developed a regression model to predict stock returns, which is the last column in our dataset.
- Calculated feature importance to identify the most influential factors affecting stock returns.
The dataset contains historical stock prices and features that potentially influence stock returns, including market capital indicators, volatility measures, and financial ratios.
stock_market_prediction.Rmd: Contains the R Markdown analysis script used for the ARIMA and LightGBM modeling.
The project successfully demonstrated the ability to predict stock returns using ARIMA and LightGBM. The feature importance analysis provided insights into the key drivers of stock performance.
The analysis can be replicated or extended by running the stock_market_prediction.Rmd script in RStudio or a similar R environment.
- R and relevant packages (forecast, lightgbm, dplyr, etc.)
- An understanding of time series analysis and machine learning concepts.
Feel free to fork the project, submit pull requests, or send suggestions to improve the models or analyses.
This project is open source and available under the MIT License.
For any additional questions or comments, please contact @jiatangzhi.
- Thanks to all the contributors and the open-source community for the tools and libraries used in this project.