/Stock-Trend-Prediction

Stock Trend Prediction is a project that leverages Long Short-Term Memory (LSTM) neural networks to predict future stock prices using historical data. By integrating advanced machine learning techniques with a user-friendly web interface, it provides insights into stock trends and forecasts.

Primary LanguageJupyter Notebook

Stock Trend Prediction

(Link to vedio - https://drive.google.com/file/d/140jeYSpt8iMTMItN4GpTYn7y23tiHjk6/view?usp=drivesdk)

This project uses a Long Short-Term Memory (LSTM) neural network to predict stock prices based on historical data. The project is implemented using Python, leveraging libraries such as TensorFlow/Keras for building the model, Streamlit for the web interface, and yfinance for fetching stock data.

Project Structure

  • app.py: Main application file that sets up the Streamlit interface, fetches data, preprocesses it, and displays the prediction results.
  • my_model.keras: Pre-trained LSTM model used for making predictions.
  • Stock Trend Prediction.ipynb: Jupyter Notebook containing the code for data processing, model training, and evaluation.

Features

  • Data Fetching: Fetches historical stock data using the yfinance library.
  • Data Visualization: Visualizes the stock's closing prices along with 100-day and 200-day moving averages.
  • Model Training: Trains an LSTM model to predict stock prices based on past data.
  • Prediction: Uses the trained model to predict future stock prices and compares them with actual prices.

Installation

Clone the repository:

git clone https://github.com/vedantl0101/stock-trend-prediction.git
cd stock-trend-prediction

Install the required packages:

pip install -r requirements.txt

Requirements:

  • keras
  • matplotlib
  • numpy
  • pandas
  • scikit-learn
  • streamlit
  • yfinance

Ensure you have my_model.keras in the root directory of the project.

Usage

To run the application, use the following command:

streamlit run app.py

This will start the Streamlit server, and you can interact with the application via your web browser.

app.py

The app.py file contains the main code for the Streamlit application. It includes:

  • User Input: Accepts a stock ticker symbol from the user.
  • Data Fetching: Fetches historical stock data for the specified ticker.
  • Data Visualization: Plots the closing prices, and 100-day and 200-day moving averages.
  • Data Preparation: Splits the data into training and testing sets and scales the data.
  • Model Loading: Loads the pre-trained LSTM model.
  • Prediction: Makes predictions on the test data and visualizes the results.

Stock Trend Prediction.ipynb

The Jupyter Notebook includes:

  • Data Processing: Loads and preprocesses the data for model training.
  • Model Building: Defines and trains the LSTM model.
  • Model Evaluation: Evaluates the model's performance on the test data.
  • Visualization: Visualizes the prediction results.

Pre-trained Model (my_model.keras)

The pre-trained LSTM model is saved as my_model.keras. This model is used in the Streamlit application to make predictions. Ensure this file is present in the root directory of the project.

Data Preparation and Model Training

  • Data Loading: Data is loaded using the yfinance library.
  • Data Splitting: The data is split into training (70%) and testing (30%) sets.
  • Scaling: The data is scaled using MinMaxScaler.
  • Model Training: The LSTM model is trained on the training data.
  • Model Saving: The trained model is saved for future use.

Visualizations

The app includes several key visualizations to help users understand stock trends and model performance:

  • Closing Price vs Time Chart

    • Description: Shows the historical closing prices of the selected stock over time.
    • Purpose: Helps users see the overall trend and fluctuations in stock prices.

  • Closing Price vs Time Chart with 100MA & 200MA

    • Description: Displays the closing price along with 100-day and 200-day moving averages.
    • Purpose: Helps users identify long-term trends and smoothing effects from moving averages.

  • Predictions vs Original

    • Description: Compares the predicted stock prices to the actual closing prices.
    • Purpose: Visualizes the accuracy of the model's predictions against real data.

  • Residual Plot

    • Description: Scatter plot showing residuals (differences between actual and predicted values) versus predicted values.
    • Purpose: Assesses how well the model's predictions match the actual values and identifies any patterns in residuals.

  • Distribution of Residuals

    • Description: Histogram displaying the distribution of residuals.
    • Purpose: Helps evaluate the spread and frequency of prediction errors, indicating model performance.