Objective: With extensive high-fidelity measured data collected from modern buildings, data science has become a promising tool for optimizing building performance and design, enhancing the built environment, and reducing environmental impacts. Data science and Machine Learning (ML) techniques have wide-ranging applications across the building life cycle, encompassing everything from the initial design phase to ongoing operation and maintenance stages. It is important to understand and practice how data science and ML technologies could enhance building design and automation, leading to more environmentally responsive and efficient buildings. The course will start from data science fundamentals, ML modeling techniques, and extend to their applications in the architecture domain.
This course aims to provide students:
1. Understanding smart and sustainable building systems, performance metrics and challenges
2. Use of computational simulation tools for data generation
3. Acquire skills for building data analysis, visualization, and system modeling using Python
4. Hands-on experience with ML model training and application in performance/design optimization
Class format: The class meets for three hours every Monday, consisting of lectures and workshops. The workshops will recap and practice what is taught during the lecture while learning coding and simulation skills required for building domain applications.
Class content: The first part of the class covers building system operations and energy flow dynamics, providing a foundational understanding of performance metrics critical to environmentally responsive buildings. The second part focuses on data collected from buildings with emphasis on techniques for data cleaning, analysis, and visualization. The third part will introduce various ML techniques for model training. Students will understand and compare the performance of different ML methods and gain the ability to find the most suitable model for a given task. In the last part, students will learn how to utilize the ML models to optimize the design and performance of environmentally responsive buildings.
Class resource: The class will utilize open-source building-related data and some available resources from HouseZero®. According to the interest and final project topics, students will have the opportunity to test their final project outcome at the HouseZero® ‘live lab’ located on the third floor.
Prerequisite: Basic understanding in environmentally responsive (a.k.a. smart, green, and high-performance) buildings and coding skills in Python will be helpful.