ayadidi/Project_ML_Energy_Load_Forecasting

An interactive Machine Learning dashboard using XGBoost, Random Forest, and Linear Regression to predict electricity consumption and manage grid load in Morocco. # Electrical Energy Consumption Prediction (Morocco) 🇲🇦 ## 📌 Project Overview This project aims to predict the electricity consumption of the city of Tetouan[cite: 4, 6, 109]. [cite_start]The goal is to provide a decision-support tool to optimize production, manage peak loads, and reduce the carbon footprint. ## 📊 Models and Performance We adopted an incremental approach by comparing three different algorithms: * **XGBoost (Final Choice):** Exceptional performance with an **R² score of 98.4%**. * **Linear Regression:** Used as a baseline, reaching **99.4%** accuracy by integrating lag variables. * **Random Forest:** Robust against non-linearities with an accuracy of **89%**. ## 🛠️ Key Influence Factors Exploratory Data Analysis (EDA) demonstrated that: * **Temperature** is the #1 determining factor due to the heavy use of air conditioning and heating. * A critical consumption peak systematically occurs at **8:00 PM (20h00)**. * The months of **July and August** show the highest electrical loads. ## 🖥️ Dashboard Application An interactive interface developed with **Streamlit** allows users to: 1. **Simulate consumption** by modifying weather parameters (Temperature, Humidity, Wind Speed). 2. **Visualize real-time predictions** through a "Traffic Light" alert system. 3. **Compare model performances** across historical data windows. ## 🚀 Installation & Setup 1. **Clone the repository:** ```bash git clone github.com cd Project_ML_Energy_Load_Forecasting