LSTM Stock Price Prediction

This repository contains code for predicting stock prices using Long Short-Term Memory (LSTM) neural networks. The project leverages technical indicators and time series data to forecast future stock prices.

Overview

The project involves the following steps:

1. Data Preprocessing: Load and clean the data, convert volumes, and set the date as the index.
2. Feature Engineering: Add technical indicators such as SMA, EMA, and RSI to the dataset.
3. Scaling: Normalize the features and target variable using MinMaxScaler.
4. Data Splitting: Split the data into training and test sets using TimeSeriesSplit.
5. Model Building: Construct and compile an LSTM model with Bidirectional LSTM layers and Dropout.
6. Model Training: Train the model using early stopping to prevent overfitting.
7. Prediction: Make predictions on the test set and future prices, ensuring the price change is within a specified range.
8. Evaluation: Evaluate the model performance using MAE, MSE, and RMSE, and visualize the results.

Getting Started

Prerequisites

Ensure you have the following libraries installed:

• pandas
• numpy
• matplotlib
• scikit-learn
• tensorflow
• ta (Technical Analysis Library)

You can install the required libraries using pip:

pip install pandas numpy matplotlib scikit-learn tensorflow ta