Renewable-Energy-Forecast

Introduction

The Spanish government is contemplating an expansion of its renewable energy sources to meet the increasing demands of the country’s electricity infrastructure. Electricity, as an essential amenity, is pivotal for fostering a meaningful standard of living and enhancing business growth. This project explores the shortfall between energy generated through fossil fuels and diverse renewable sources in Spain. The research aims to yield actionable, data-driven insights empowering the government to plan and manage energy supply efficiently. This, in turn, ensures a reliable and sustainable energy infrastructure while curbing carbon emissions.

Goal of the Project

The project’s overarching objective is to train and deploy a regression model addressing a critical issue in Spain’s energy sector. By predicting the shortfall between energy generated from fossil fuels and renewable sources, the model facilitates effective planning for Spain’s transition to renewable energy. This ensures a reliable and sustainable energy supply while mitigating carbon emissions from non-renewable sources.

Outcome

We developed a robust regression model capable of predicting the shortfall between energy generated by fossil fuels and renewable sources. We trained the model leveraging supervised machine learning techniques, including linear regression, random forests, decision trees, and LGBM. Following a meticulous evaluation using root mean square error (RMSE) as the metric, the LGBM emerged as the best-performing model and was subsequently adopted and deployed. Furthermore, the model excelled in a Kaggle competition, securing the first position.

Skills and Tools Used

• Python
• Machine learning packages (Scikit-learn and nltk)
• Project Management: Trello Board
• Data Cleaning
• Data Exploration
• Model Deployment: AWS EC2 and Flask
• Soft Skills: Presentation skill, Collaboration, Communication and Leadership

Task Overview (My contribution)

As the technical lead for the project, here is a high-level overview of my contribution:

• Conducted comprehensive data exploration, identifying and rectifying data quality and structural issues to ensure the integrity of the dataset.
• Designed, planned, and prioritized tasks, ensuring a seamless workflow and timely project execution.
• Led the design and experimentation of models and techniques to optimize algorithm selection for problem modeling.
• Authored and reviewed code, ensuring adherence to standard practices.
• Presented technical solutions and model results to both technical and non-technical audiences.

Modeling

Among the regression models trained, including Linear Regression, Decision Trees, Light GBM (LGBM), Random Forest, and Support Vector Machine, LGBM gave the best RMSE score.

Insights

Through comprehensive data exploration and analysis, some of the hidden patterns emerged:

• Daily Variation: A higher shortfall is observed between 9 and 12 hours, corresponding to office hours and periods of increased domestic gadget usage for house chores. Addressing this peak demand period is imperative for the government.
• Weekly Variation: Weekdays exhibit a higher load shortfall than weekends, emphasizing electricity’s critical role in daily work routines.
• Seasonal Variation: The second half of the year, corresponding to summer and winter, witnesses a high shortfall. This signifies increased electricity consumption for activities such as increased use of cooling appliances.

Recommendations

To enhance renewable energy production in Spain, we propose the following recommendations:

• Consider integrating energy storage solutions to mitigate shortfalls during peak usage hours.
• Implement strategies to boost renewable energy generation on weekdays, addressing increased demand during working hours.
• Encourage energy-efficient practices and raise awareness to achieve a more balanced load distribution during the week.
• Promote the adoption of energy-efficient technologies for cooling and heating purposes to reduce overall demand during peak seasons.

Acknowledgements

My Shout out goes to my team members for their contributions:

• Jacinta Muindi
• Abiodun Adeagbo
• Seye Tare Ganranwei
• Pere Ganranwei
• Job Onose Okhueleigbe

Ready to unlock the full potential of your data? I’m excited to collaborate and help your business thrive through data-driven insights. Get in touch!