📈 http://ishai.pro
Hello, and welcome to my latest project! I’m excited to share a new tool I’ve developed: a Python-based stock prediction application powered by a Long Short-Term Memory (LSTM) neural network. This project isn’t just about code; it’s about making a powerful deep learning model accessible to anyone interested in analyzing stock market trends.
The goal was to create a solution that is both accurate and easy to use. I accomplished this by integrating three key technologies:
1. The Brains: PyTorch and an LSTM Model ðŸ§
At the heart of this application is a custom-built LSTM model, which is a type of recurrent neural network (RNN) especially good at processing sequences of data, like stock prices over time. Unlike simpler models, LSTMs can “remember” long-term dependencies in the data, making them a great fit for time series analysis.
My script uses the PyTorch library to train this model. PyTorch is a flexible and powerful deep learning framework that allows for efficient computation, especially when leveraging a GPU for training. My model is designed with multiple layers and a dropout mechanism to prevent overfitting, ensuring it can make more reliable predictions on new, unseen data.
2. The Data: yfinance 💰
To make predictions, the model needs a continuous flow of high-quality data. I used the yfinance library to pull historical stock data directly from Yahoo Finance. This allows the application to be dynamic—you can enter a company’s ticker symbol, and the script will automatically fetch the past five years of its closing prices to train the model. This is a crucial step that ensures the model’s predictions are based on the most recent market trends.
3. The Interface: Streamlit ✨
What good is a powerful script if it’s hard to use? I chose Streamlit to build a clean and interactive user interface. This is what transforms the Python script into a fully functional web application. Streamlit handles everything from the user input for the stock ticker to displaying the results in a clear and visual way.
Within the app, you can:
- Enter any stock ticker from a pre-defined list.
- Watch as the model trains and makes predictions.
- View a side-by-side comparison of the actual vs. predicted stock prices on a graph.
- See key performance metrics like RMSE, MAE, and R² score, which tell you how well the model performed.
- Get a look into the future with a 30-day price prediction graph.
This combination of PyTorch, yfinance, and Streamlit allows for a seamless experience that demonstrates the power of machine learning in a practical and engaging way.