AirQNet (Air Quality Network)

AirQNet is a full end-to-end PyTorch-based project for air quality classification—covering data ingestion, preprocessing, model training, evaluation, and inference. This repository demonstrates the entire AI modeling pipeline using the Air Quality and Pollution Assessment dataset.

• Model Size: 10,886 parameters
• Accuracy: ~93% on the test set
• Framework: PyTorch with additional libraries like NumPy, Pandas, and scikit-learn
• Key Features:
  • Data loading and feature normalization
  • Feedforward neural network (FFN) with BatchNorm and Dropout
  • Cross-entropy loss for multi-class classification
  • AdamW optimizer with weight decay
  • Early stopping and automatic checkpointing
  • Live plotting of loss and accuracy during training
  • Inference script for single and bulk prediction

---

Datasets

• Air Quality and Pollution Assessment
• The repository includes scripts to split and preprocess the CSV data.

---

Training and Evaluation

1. Training Script (main.py)

   • Loads the dataset via train_loader / val_loader.
   • Initializes the model FFN, optimizer, and hyperparameters.
   • Implements early stopping to halt training when validation loss stops improving.
   • Visualizes training & validation metrics in real time.

2. Evaluation Script (evaluation.py)

   • Loads the trained model from a .pth file.
   • Computes accuracy and class probabilities on a test set.
   • Provides utility to predict individual samples or entire CSV files.
   • Exports predicted classes and probabilities to a CSV.

---

Results

Plots

---

Model Computational Graph (torchlens)

Usage

1. Install Dependencies

   pip install -r requirements.txt

2. Train the Model

   python main.py

3. Evaluate / Predict

   python evaluation.py