Building Data Analysis

Zixiao (Shawn) Shi

Email: [email protected] or [email protected]

This class introduces fundamental topics on analyzing building energy data. It focuses on building energy data but the course contents are generalizable to other data analytic topics. This class has been taught as ENVE5701/CIVE5704 at the Carleton University, Ottawa Canada. It covers topics such as descriptive statistics, visualization, hypothesis testing, clustering, regression, forecasting, classification and causal inference. Students are expected to have a basic understanding of statistics, building science and computer programming (preferrably Python).

Links to Lecture notes and Jupyter Notebooks are hosted below.

Note class material may be updated intermittently. Any feedback welcomed!

Lecture Slides and Workshop Notebooks

• Slides: Introduction
• Slides: Descriptive Statistics Slides
  • Jupyter Notebook: Descriptive Statistics with NYC Building Energy Benchmark Data
• Slides: Probability Basics
• Slides: Data Visualization
  • Jupyter Notebook: Visualization with NYC Building Energy Benchmark Data
• Slides: Hypothesis Testing
  • Jupyter Notebook: Hypothesis Tests with Seattle Building Energy Benchmark Data
• Slides: Timeseries Characteristics
  • Jupyter Notebook Time Series Manipulation with ASHRAE Great Energy Predictor/Building Genome II Dataset using Pandas
• Slides: Intro to Machine Learning, Clustering
  • Jupyter Notebook: Clustering of Time Series Energy Data
• Slides: Regression, part 1
  • Jupyter Notebook: Linear Regression as Energy Benchmark Model
• Slides: Regression, part 2
  • Jupyter Notebook: Advanced Regression as Energy Benchmark Model
• Slides: Forecasting
  • Jupyter Notebook: Forecasting Building Energy Consumption
• Slides: Classification
  • Jupyter Notebook: Classification of Building Types with Seattle Building Energy Benchmark Data
• Slides: Outlier Analysis
  • Jupyter Notebook: Outlier Detection of Time Series Energy Data
• Slides: Causal Inference
• Slides: Review