Stock Price Prediction Project

Overview

This project focuses on predicting stock prices using machine learning techniques, primarily Long Short-Term Memory (LSTM) models, and compares their performance with traditional regression-based models. The dataset includes historical stock price data for multiple companies, which is preprocessed and explored to gain insights into trends and patterns.

Methodology

Data Collection and Preprocessing

• We collected historical stock price data for multiple companies from the provided datasets.
• Data preprocessing steps included renaming columns, converting date formats, dropping unnecessary columns (e.g., 'open_interest'), handling missing values, and scaling features.

Exploratory Data Analysis (EDA)

• We visualized the closing prices of each company to gain insights into historical trends and patterns.

Feature Engineering

• We created additional features, such as moving averages, to provide context and capture underlying patterns in the data.

Model Building

LSTM Models

• We constructed two LSTM architectures: LSTM1 and LSTM2, with different configurations of LSTM layers and batch sizes.
• Hyperparameters such as learning rate, batch size, and number of epochs were fine-tuned to optimize model performance.
• The models were trained using the Adam optimizer and evaluated using metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE).

Traditional Regression Models

• We also trained traditional regression-based models, including Linear Regression and XGBoost Regressor, for comparison with LSTM models.
• These models were trained using the same dataset and evaluated using the same metrics.

Model Evaluation

• We evaluated the performance of each model using metrics such as RMSE, MSE, and MAE on both training and validation sets.
• The models were also tested on unseen data (testing set) to assess their generalization ability.

Results and Analysis

• LSTM models demonstrated competitive performance compared to traditional regression-based models.
• LSTM2, with a time_series_steps of 30, batch_size of 128, and 10 epochs, showed the best performance in terms of prediction accuracy.
• The LSTM models outperformed Linear Regression and XGBoost Regressor in terms of RMSE, MSE, and MAE metrics.
• Visualization of predicted vs. actual stock prices showed that LSTM models accurately captured the underlying patterns in the data.

Conclusion

• LSTM models offer a promising approach for stock price prediction, leveraging their ability to capture long-term dependencies in sequential data.
• Fine-tuning hyperparameters such as batch size and number of epochs is crucial for optimizing LSTM model performance.
• While traditional regression-based models can also provide reasonable predictions, LSTM models tend to offer better accuracy, especially for time-series forecasting tasks.

Future Work

• Further experimentation with different LSTM architectures and hyperparameters could potentially improve model performance.
• Ensemble methods, such as combining LSTM models with traditional regression models, could be explored to leverage the strengths of both approaches.
• Integration of additional data sources, such as external market indicators or news sentiment analysis, may enhance the predictive capabilities of the models.

Code and Documentation

For the full project code and documentation, please visit the GitHub repository.