Course: Machine Learning · Dr. Shazia Saqib
Live URL: https://volleyball-whale-res-within.trycloudflare.com
A full-stack ML application from data pipeline to interactive UI
Random Forest and XGBoost models predict household electricity load (kW) from 30 weather and appliance features. Test R² ≈ 0.76.
Real-time weather from Open-Meteo for 18 Pakistani cities (Islamabad, Lahore, Karachi, etc.) feeds the model for instant predictions.
House load gauge in real Amperes at Pakistan's standard 230 V single-phase, with sanctioned-load threshold and trip-risk alerts.
9-tier Pakistani domestic tariff with FPA, QTA, GST, electricity duty, PTV fee, meter rent. Generate downloadable PDF bills.
Decides which appliances to keep ON or shut OFF based on weather, hour, peak/night windows. Estimates kWh and PKR savings.
540 bills across 90 households, all 11 DISCOs (IESCO, LESCO, KE, FESCO, MEPCO, GEPCO, PESCO, QESCO, HESCO, TESCO, SEPCO).
Search by Reference Number or CNIC, view full billing history, consumption trends, unpaid bills, generate any month's PDF.
Real-time notifications when load exceeds sanctioned amperes, monthly target, slab thresholds, or extreme temperature events.
Production-grade Python + ML + cloud delivery
numpy 2.4 · pandas 3.0 · scikit-learn 1.9 · XGBoost 3.2
Interactive dashboard with animated CSS, glass cards, gauge charts, and smart layout.
Server-side PDF generation for downloadable Pakistani-format electricity bills.
This page on Cloudflare Pages. The live ML app on Cloudflare Tunnel — both edge-routed.