Air Quality Forecaster - GRU Neural Network

📂 Active Dataset

📊 Current Air Quality

--

AQI

Loading...

PM2.5: -- µg/m³

Last updated: --

💼 Business Value & Impact Analysis

🎯 Real-World Applications: This AQI forecasting system delivers measurable business value across multiple sectors through predictive analytics and proactive decision-making.

🏥

Public Health Protection

--

Proactive health advisories for at-risk populations

0 high-risk hours detected in next 24h

💰

Cost Savings Potential

$0

Estimated healthcare & operational savings

Based on predictive maintenance & health interventions

⚙️

Operational Efficiency

--

Smart scheduling & resource optimization

0 optimal hours for outdoor activities

🛡️

Risk Mitigation

--

Early warning system for pollution events

0 hours advance notice

📊 ROI Impact Calculator

Population Affected:

Avg Healthcare Cost/Person:

Prevention Rate: %

🏥 Health Advisory & Recommendations

⚠️ Current Risk Level

📋 Recommended Actions

👥 Vulnerable Groups

⏰ 24-Hour Risk Timeline

🔮 24-Hour GRU-Based Future Forecast

🧠 AI-Powered Prediction: Using Gated Recurrent Unit (GRU) neural network to analyze temporal patterns and predict future air quality levels.

📅 Extended GRU Future Prediction (7-Day Outlook)

🔮 Long-term Forecast: Multi-day air quality predictions using iterative GRU forecasting to project trends up to 7 days ahead.

Forecast Duration:

📈 Historical Data

Time Range:

⏰ Hourly Pattern Analysis

🌍 Country-wise Air Quality Analysis

📊 Geographic Distribution: Analyzing air quality patterns across different countries in your uploaded dataset.

📤 Upload Custom Dataset

Upload your own air quality CSV file to analyze and forecast.

Required columns: Any column with "date" or "time" + Any column with "PM", "AQI", or "pollution"

Examples: Timestamp,PM2.5 | Date,AQI Value | datetime,pollution

💡 Tip: Open browser console (F12) to see detailed parsing logs

❓ Need help with CSV format?

Quick CSV Template:

Timestamp,PM2.5
2025-01-01 00:00:00,25.3
2025-01-01 01:00:00,23.7
2025-01-01 02:00:00,21.2

Common Issues:

❌ Missing header row (column names)
❌ Column names don't contain "date/time" or "PM"
❌ Only 1-2 rows of data (need at least 24 for good forecasts)
❌ PM2.5 values are text instead of numbers

✅ Try the sample file first: sample_data.csv in this folder

🧠 About the Model

GRU Neural Network

This application uses a Gated Recurrent Unit (GRU) neural network to forecast air quality levels. GRU is a type of recurrent neural network that excels at learning patterns in time-series data.

Model Architecture

Input: 24 timesteps × 11 features
GRU Layer 1: 64 units with Dropout & BatchNorm
GRU Layer 2: 32 units with Dropout & BatchNorm
Dense Layers: 32 → 16 → 1
Output: PM2.5 prediction

Features Used

PM2.5 levels (target variable)
Cyclical time encoding (hour, day, month)
Weekend indicator
Rolling averages (3h, 6h)
Rolling standard deviation (3h)

🌬️ Air Quality Forecaster