📈 Time Series Forecasting: LSTM & XGBoost

📖 Overview

This project focuses on forecasting time-series data (specifically electricity load/consumption) using machine learning and deep learning techniques. We compare the performance of gradient boosting models (XGBoost) against deep recurrent neural networks (LSTM) to predict future values based on historical patterns.

The project covers the full data science lifecycle:

1. Data Preprocessing: Cleaning, scaling, and handling missing values.
2. Exploratory Data Analysis (EDA): Visualizing trends, seasonality, and stationarity.
3. Model Development: Training XGBoost and LSTM models.
4. Evaluation: Comparing models using RMSE and MAE.

---

📂 Project Structure

├── Scripts-Client/
│   ├── Preprocessing.ipynb   # Data cleaning, scaling, and formatting
│   ├── Eda.ipynb             # Exploratory Data Analysis & Visualization
│   ├── LSTM.ipynb            # Deep Learning Model (Long Short-Term Memory)
│   ├── Xgboost.ipynb         # Gradient Boosting Model
│   └── Splitting_data.py     # Helper script for train/test splits
├── Models/
│   ├── lstm_model.keras      # Saved LSTM model
│   ├── XGB_mode.pkl          # Saved XGBoost model
│   └── Sclar.pkl             # Saved Scaler object (for inverse transform)
└── README.md