---

📈 Stock Price Prediction with LSTM (Streamlit App)

A Streamlit-based web application that predicts stock prices using a Long Short-Term Memory (LSTM) neural network. Users can upload historical stock price data, tune model hyperparameters interactively, train the model, and visualize actual vs predicted prices along with performance metrics.

---

📑 Table of Contents

• Overview
• Architecture
• Pipeline
• Features
• Tech Stack
• Requirements
• Installation
• How to Run
• Usage Guide
• Model Architecture
• Evaluation Metrics
• Important Notes
• Future Enhancements
• License

---

🔍 Overview

This application demonstrates how LSTM-based deep learning models can be applied to time-series forecasting in finance. It allows end users to:

• Upload custom stock datasets
• Experiment with different hyperparameters
• Train an LSTM model in real time
• Visualize predictions and training performance

---

🏗 Architecture

---

🔄 Pipeline

---

✨ Features

• 📂 CSV Upload – Upload historical stock price datasets

• 👀 Data Preview – Inspect uploaded data before training

• 🎛 Dynamic Hyperparameter Tuning

  • Sequence length (look-back window)
  • Epochs
  • Batch size
  • LSTM units (Layer 1 & 2)
  • Train–test split ratio

• 🚀 One-click Model Training

• 📉 Training Loss Visualization

• 📊 Performance Metrics – Mean Squared Error (MSE)

• 📈 Interactive Prediction Plot – Actual vs Predicted prices

---

🛠 Tech Stack

• Frontend: Streamlit
• Backend / ML: TensorFlow, Keras (LSTM)
• Data Processing: NumPy, Pandas
• Visualization: Matplotlib
• Preprocessing: Scikit-learn (MinMaxScaler)

---

📦 Requirements

• Python 3.8+

• Libraries:

  streamlit
  numpy
  pandas
  matplotlib
  scikit-learn
  tensorflow

---

⚙️ Installation

pip install streamlit numpy pandas matplotlib scikit-learn tensorflow

---

▶️ How to Run

1. Save the Application

   app.py

2. Prepare CSV Data Your dataset must contain a Close column.

   Example format:

   Date,Open,High,Low,Close,Volume
   2023-01-01,100.0,102.5,99.5,101.0,100000
   2023-01-02,101.5,103.0,100.0,102.0,120000

3. Run the App

   streamlit run app.py

4. Access the Application

   http://localhost:8501
   

---

🧭 Usage Guide

1. Upload CSV via the file uploader

2. Verify Data Preview (ensure Close column exists)

3. Configure Hyperparameters using sidebar sliders:

   • Look-back window
   • Epochs
   • Batch size
   • LSTM units
   • Train–test split

4. Train Model using the Train Model button

5. Analyze Results:

   • Training loss curve
   • Mean Squared Error (MSE)
   • Actual vs Predicted stock price graph

---

🧠 Model Architecture

The model is built using Keras Sequential API:

• MinMaxScaler Normalizes prices between 0 and 1

• Sequence Generator (create_sequences) Converts time series data into LSTM-ready sequences

• LSTM Layer 1

  LSTM(units=lstm_units_1, return_sequences=True)

• LSTM Layer 2

  LSTM(units=lstm_units_2)

• Dense Output Layer

  Dense(1)

• Optimizer: Adam

• Loss Function: Mean Squared Error

---

📊 Evaluation Metrics

• Mean Squared Error (MSE) Measures average squared difference between actual and predicted prices

---

⚠️ Important Notes

• 🚨 For educational purposes only

• 📉 Stock market prediction is highly volatile and uncertain

• 📊 Model performance depends heavily on:

  • Data quality
  • Data size
  • Hyperparameter choices

• ❌ Not intended for financial or investment advice

---

🚀 Future Enhancements

• Add technical indicators (RSI, MACD, EMA)
• Support multi-stock prediction
• Integrate real-time data via APIs
• Deploy as a cloud-based API (GCP / AWS)
• Add model comparison (LSTM vs GRU vs Transformer)

---

📄 License

This project is licensed under the MIT License.

---