🌦️ Weather Prediction using Bayesian Networks and PyTorch

This project predicts weather parameters (like humidity, rainfall, etc.) using both Bayesian Networks (pgmpy) and Neural Networks (PyTorch). It includes MLflow for experiment tracking and Prometheus for monitoring.

---

📁 Dataset

• Name: weatherAUS.csv
• Source: Kaggle - Australian Weather Dataset

---

📌 Features Used

• MinTemp, MaxTemp, Rainfall, WindGustSpeed, WindSpeed9am, WindSpeed3pm
• Humidity9am, Humidity3pm, Pressure9am, Pressure3pm
• Temp9am, Temp3pm, RainToday, RainTomorrow

---

🧠 Models Implemented

1. Bayesian Network (via pgmpy)

• Learns structure using HillClimbSearch
• Parameters via MaximumLikelihoodEstimator
• Predicts missing values

2. PyTorch Neural Network

• Multi-layer Feedforward Network
• Hyperparameter tuning using K-Fold Cross-Validation
• Tracks best hyperparameters and MSE

---

🚀 How to Run

pip install -r requirements.txt

python newBayesian.py

mlflow ui --port 5000

---

Visualizations

Correlation Matrix

Feature Distributions

Prediction vs Actual Plots

Error Histograms

---

Tech Stack

Python 🐍

PyTorch 🔥

pgmpy 📊

MLflow 🧪

Prometheus 📡

Pandas, NumPy, Seaborn, Scikit-Learn

---

Fork

feel free to fork and made changes to this code

---