title	emoji	colorFrom	colorTo	sdk	sdk_version	app_file	pinned
Dogecoin LSTM Predictor	🐶	yellow	red	streamlit	1.35.0	app.py	false

🐶📈 Dogecoin Price Predictor (LSTM-based)

Get ready to predict Dogecoin prices with the power of LSTM!

Dive into the world of cryptocurrency forecasting with this deep learning project! Using LSTM (Long Short-Term Memory) networks, this app predicts Dogecoin (DOGE) prices by analyzing historical data. Whether you're a crypto enthusiast or a data science learner, this project offers a blend of sequence modeling, time series forecasting, and interactive visualization.

🚀 Project Overview

• 📊 Input Data: Historical DOGE-USD prices
• 🤖 Model Type: LSTM Neural Network (with Keras & TensorFlow backend)
• 🧠 Features: Closing price series (scaled), time-based sequence modeling
• 🎯 Objective: Predict future closing prices of Dogecoin
• 📉 Visualizations: Loss curve, actual vs predicted, residual diagnostics, future price projection

---

📌 Highlights

• ✅ Built and trained a custom LSTM network on preprocessed crypto time series
• ✅ Achieved strong performance on test set (R² score, low residuals)
• ✅ Visualized residual distribution, model diagnostics, and future forecasts
• ✅ Saved model and scaler for real-time or batch inference
• ✅ Deployed on Hugging Face using Streamlit

---

🌐 Live Demo on Huggingface 🤗

👉 Try the App on Hugging Face Spaces

🧰 Tech Stack

Tool/Library	Purpose
Python (3.12)	Core programming language
TensorFlow/Keras	LSTM Model Building & Training
Scikit-learn	Metrics, preprocessing
Pandas, NumPy	Data manipulation
Matplotlib, Seaborn	Plotting & visualization
Streamlit	Web app deployment
Git/GitHub	Version control & collaboration

---

📂 Project Structure

📁 Dogecoin-LSTM-Predictor/
│
├── DOGE-USD.csv                   # Raw historical Dogecoin data
├── dogecoin_lstm.ipynb           # Main Jupyter Notebook (core logic)
├── lstm_stock_model.keras        # Trained LSTM model (Keras format)
├── scaler.save                   # Saved MinMaxScaler for future inference
│
├── actual_vs_predicted.png       # Visualization of model performance
├── residual_distribution.png     # Histogram + KDE of residuals
├── residual_diagnostics.png      # Residuals vs predicted plot
├── loss_curve.png                # Training loss progression
├── future_price_prediction.png   # Future forecasted prices
│
└── README.md                     # This file ✨

---

📈 Sample Results

Metric	Value
R² Score (Test)	~0.90+
MAE / MSE	Low (well-optimized)

---

🧪 How to Run

# 1. Clone the repo
git clone https://github.com/Kurra-Srinivas/Dogecoin-LSTM-Predictor.git
cd Dogecoin-LSTM-Predictor

# 2. (Optional) Create virtual environment
python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows

# 3. Install requirements
pip install -r requirements.txt

# 4. Run the Streamlit app
streamlit run app.py

---

🙋‍♂️ Author

Kurra Srinivas
📧 Email: srinivaskurra886@gmail.com
🔗 LinkedIn: https://www.linkedin.com/in/kurra-srinivas-31727420b/
🐙 GitHub: Kurra-Srinivas