ALDI++: Automatic and parameter-less discord and outlier detection for building energy load profiles

  title={ALDI++: Automatic and parameter-less discord and outlier detection for building energy load profiles},
  author={Matias Quintana and Till Stoeckmann and June Young Park and Marian Turowski and Veit Hagenmeyer and Clayton Miller},
Data-driven building energy prediction is an integral part of the process for measurement and verification, building benchmarking, and building-to-grid interaction. The ASHRAE Great Energy Predictor III (GEPIII) machine learning competition used an extensive meter data set to crowdsource the most accurate machine learning workflow for whole building energy prediction. A significant component of the winning solutions was the pre-processing phase to remove anomalous training data. Contemporary… 

Figures and Tables from this paper


Limitations of machine learning for building energy prediction: ASHRAE Great Energy Predictor III Kaggle competition error analysis
This analysis reveals the limitations for machine learning using the standard model inputs of historical meter, weather, and basic building metadata and forms the foundation for suggestions to reduce machine learning errors by collecting and using additional training data from onsite and web-based sources to improve the capability, accuracy, scalability, and usability of machine learning.
Learning Dominant Usage from Anomaly Patterns in Building Energy Traces
This two-fold approach first leverages the Matrix Profile technique for time series data mining to build a dataset of anomaly patterns from public building energy traces and extract analytics information, and uses the labeled dataset in a supervised learning classification model to discriminate between various related dominant usage patterns.
Data driven parallel prediction of building energy consumption using generative adversarial nets