Initial Release

🎉 v1.0.0 - Initial Release

---

📌 Overview

The first official release of the Time Series Analysis Using Machine Learning project! This version focuses on predicting heavy metal concentrations in industrial wastewater using advanced models such as ARIMA and PSO-LSTM. The release includes robust data preprocessing, sensitivity analysis, and performance evaluation to optimize wastewater treatment processes.

---

🚀 Features

1. ARIMA Model

• Implements ARIMA for time series forecasting.
• Grid search for optimal parameters (p, d, q).
• Performance Metrics: MSE, MAE, R².

2. PSO-LSTM Model

• Combines Particle Swarm Optimization (PSO) with Long Short-Term Memory (LSTM) networks.
• Optimizes hyperparameters such as learning rate, dropout ratio, and batch size.
• Performance Metrics: Training/Validation Loss, MSE, MAE.

3. Sensitivity Analysis

• Identifies key features like Electrical Conductivity and Chemical A as critical predictors.
• Visualizes feature importance through MSE changes.

4. Evaluation Metrics

• Provides comprehensive metrics: MSE, MAE, MSLE, and R².
• Includes a comparison between ARIMA and PSO-LSTM models.

5. Visualization

• Plots include:
  • Sensitivity analysis results.
  • ARIMA vs. LSTM comparison.
  • Model training and validation loss curves.

---

📂 Assets

Source Code

• Full project code is available for download.

Pre-trained Models

• ARIMA: models/arima_model.pkl
• PSO-LSTM: models/pso_lstm_model.h5

Sample Dataset

• Sample Data: data/sample_data.csv (a small dataset for testing).

---

📊 Performance Summary

Model	MSE	MAE	R²
ARIMA	0.053	0.025	90%
PSO-LSTM	0.021	0.064	85%

---

📝 Known Issues

• Original raw dataset is not included due to confidentiality agreements.
• Current version focuses on ARIMA and PSO-LSTM. Hybrid models (e.g., ARIMA-LSTM) will be added in future versions.

---

📅 What’s Next

• Develop a hybrid ARIMA-LSTM model to leverage the strengths of both approaches.
• Add real-time monitoring features for practical deployment.
• Enhance hyperparameter tuning using advanced optimization methods.

---

💾 Downloads

• Source Code (ZIP)
• Pre-trained Models

---