Energy Price Forecasting - Random Forest Regressor Algorithm

This repository contains a conceptual example of forecasting energy market prices using machine learning techniques. The project demonstrates data preprocessing, feature engineering, and price prediction using a Random Forest Regressor, along with result visualization for comparison between actual and predicted prices.

Note: The dataset used in this project (market_data.csv) contains confidential energy market information and cannot be shared publicly. Therefore, this repository includes only the source code, not the dataset.

---

📘 Project Overview

Energy price forecasting is crucial for energy traders, utilities, and policymakers to make informed operational and strategic decisions. This project showcases how time series data (such as demand, renewable generation, and historical prices) can be used to build a predictive model for short-term energy price forecasting.

---

🧠 Workflow Summary

1. Data Loading

The project expects a CSV file named market_data.csv containing historical data with columns such as:

• Date – Timestamp of the observation
• Price – Historical market price of energy
• Demand – Energy consumption or demand data
• Renewable_Generation – Contribution from renewable sources

⚠️ Since the data is private, the code includes a safety check to prevent execution if the file is not found.

2. Feature Engineering

• Converts Date column to datetime format and sets it as the index.
• Creates lagged features (e.g., previous day’s price) to capture time dependencies.
• Removes missing values to ensure model stability.

3. Model Definition

• Uses a Random Forest Regressor from scikit-learn as a baseline predictive model.
• Features: Price_lag1, Demand, Renewable_Generation
• Target: Price

4. Training and Testing

• Splits the dataset into 80% training and 20% testing sets (time-based, no shuffling).
• Fits the model on training data and evaluates it on test data.

5. Prediction and Visualization

• Predicts energy prices for the test set.
• Visualizes actual vs predicted prices using matplotlib to assess performance visually.

---

📊 Example Visualization

The script produces a plot comparing actual and predicted prices over time:

Energy Price Forecasting
│
├── Actual Prices (Solid Line)
└── Predicted Prices (Dashed Line)

This helps assess how well the model captures short-term market movements.

---

🛠️ Technologies Used

• Python 3.x
• Pandas – Data manipulation and preprocessing
• Scikit-learn – Machine learning and model training
• Matplotlib – Data visualization

---

⚙️ How to Use

1. Clone this repository:

   git clone https://github.com/yourusername/Energy-Price-Forecasting.git
   cd Energy-Price-Forecasting

2. Install dependencies:

   pip install -r requirements.txt

3. Provide your own dataset: Save your CSV file as market_data.csv in the project directory with the following columns:

   Date, Price, Demand, Renewable_Generation
   

4. Run the notebook:

   Open Energy Price Forecasting_ML.ipynb in Jupyter and run all cells:

   jupyter notebook "Energy Price Forecasting_ML.ipynb"

---

🚫 Data Availability

Important: The dataset used in this analysis is confidential and cannot be shared. This repository is intended for educational and demonstration purposes, showcasing the methodology and workflow rather than the actual data.

---

📊 Model Evaluation

The notebook prints quantitative metrics after prediction:

Metric	Description
MAE	Mean Absolute Error — average magnitude of errors
RMSE	Root Mean Squared Error — penalises large errors more
R²	Coefficient of Determination — proportion of variance explained

---

🔧 Changelog

Code Review Fixes (2026-04-04)

The following issues were identified and resolved:

#	Issue	Fix
1	exit() crashes Jupyter kernel	Replaced with raise FileNotFoundError(...) for proper exception handling
2	KeyError: 'Date' when CSV has whitespace in column headers	Added data.columns = data.columns.str.strip() to normalize column names
3	No validation of required columns before processing	Added explicit column presence check with a descriptive ValueError
4	No model evaluation metrics	Added MAE, RMSE, and R² score output after prediction
5	Missing requirements.txt	Added requirements.txt listing all dependencies
6	README referenced non-existent forecast_energy_price.py	Updated to reference the correct .ipynb notebook file
7	Missing numpy import (needed for np.sqrt)	Added import numpy as np

---

📈 Future Enhancements

• Integrate LSTM or XGBoost for improved time-series forecasting accuracy.
• Add MAPE (Mean Absolute Percentage Error) for business-friendly reporting.
• Add hyperparameter tuning for optimal model performance.
• Deploy as a web dashboard using Streamlit or Dash for real-time prediction visualization.

---