This event has been developed and hosted by the CIB Student Chapter of the Politecnico di Milano.
Objective
This bootcamp aims to equip participants with the necessary knowledge and practical skills to leverage ML techniques for analyzing the energy performance of buildings and clustering them based on energy-related parameters. It will provide a solid foundation in ML concepts and implementation phases, data preprocessing and cleaning, clustering techniques, and energy performance analysis and modeling. By the end of the bootcamp, participants will be able to apply ML techniques, using Python, to analyze building energy performance certificate (EPC) data, identify patterns, cluster buildings based on their energy performance, and predict their energy saving potentials if retrofitted.
Contribution
This extensive bootcamp contributes to the body of knowledge of two main cutting edge and important disciplines:
- Digitalization and automation of decision-making processes in the construction sector using the advancements of AI and ML
- Energy retrofit of existing buildings to decrease their energy consumption and carbon emission levels and contribute to the sustainability of the building sector
Audience
The event is aimed at students, researchers, data analysts, policy makers, consultants and practitioners in the construction, informatics, and energy sectors, who want to learn about the interdisciplinary ML-based solutions in a practical and case-based manner.
Agenda
Session 1: Data Preprocessing for Building Energy Analysis
- Introduction to machine learning and its applications in the building industry
- Basics of Python programming language and relevant libraries (e.g., NumPy, Pandas, Matplotlib)
- Energy performance of buildings and energy performance certificates
- Accessing and using the national and international building energy
- performance certificates databases
- Cleaning and preprocessing of building energy data
- Handling missing data, errors, and outliers
- Data graphical presentation and plotting techniques
Session 2: Unsupervised Learning Techniques for Building Clustering
- Overview of unsupervised learning algorithms
- Feature selection and Dimensionality reduction techniques (e.g., PCA) for clustering
- Implementation of various algorithms to cluster buildings based on their energy labels, construction periods, and building components’ thermal transmittance.
- Evaluation metrics for clustering
- Graphical presentation of building cluster for analysis
Session 3: Machine Learning Modeling for Energy Performance Analysis and energy saving prediction
- Introduction to ML algorithms and Neural Networks for regression to predict the energy consumption of buildings
- Feature engineering for energy performance analysis
- Evaluation and Validation of the model
- Comparative analysis of various ML algorithms’ performance
- Discussion and feedback session
- Recap of key learnings and next steps
Requirements
To participate effectively in the bootcamp you will require:
- Basic Python programming knowledge
- Familiarity with fundamental concepts of ML
- A laptop with Python, Google Colab, and relevant libraries installed (instructions will be communicated through email before the main session)
Logistics
The virtual bootcamp will be held on October 11, 18, and 25 2023, 17:00-19:00 CET, and will consist of 3 sessions of 2 hours each. There is no charge to attend.
Instructors
The bootcamp will be instructed by Professor Fulvio Re Cecconi, an Associate Professor in the Department of Architecture, Built Environment, and Construction Engineering at Politecnico di Milano and Ania Khodabakhshian, a PhD Candidate at the same department.
