Scale-to-Zero Retail Forecasting System

إليك وصف العمل الخاص بمشروع التنبؤ بالتجزئة (Scale-to-Zero Retail Forecasting System - M5 Big Data Capstone) باللغة الإنجليزية وجاهز للنسخ المباشر:

* **Forecasting System:** Engineered a production-grade forecasting system handling 30k+ streams across 58M sparse rows using an ARIMA forecasting model.

* **Big Data Optimization:** Optimized Big Data storage converting CSVs to compressed formats and PySpark schema tuning reducing memory usage by 60-75%.

* **Architecture & Deployment:** Built local FastAPI + React stack (<35ms latency) and Azure Cloud "Scale-to-Zero" architecture (1.77 RMSE, 97.1% accuracy).

* **Technologies Used:** Python, PySpark, ARIMA, FastAPI, React, Azure Cloud, Git, GitHub.