The BIGG project is assessing the potential of big data for building energy efficiency enhancement. Through 6 business cases addressing most of the common challenges faced by the energy efficiency industry, the project aims at providing a solution kit to enable data analytics to increase European buildings’ level of performance.

As part of this solution kit, the BIGG consortium is working on the design of an AI toolbox to provide energy managers and energy analysts with a set of AI functions developed specifically for the fulfillment of these business cases. After the project crossed the one-year line, the AI toolbox is now entering the implementation phase.

With a specific focus on the needs of the identified business cases, the BIGG AI toolbox is proposing a set of functions enabling existing platforms to push their analytics capabilities.

The functions developed are articulated across 4 main sections.

Data preparation

Timestamp management
Data Outlier detection
Missing data management
Data validation

Data transformation

Data Profiling
Weather data management
Autoregressive processes
Calendar component management
Cyclic time series management
Non-routine adjustments


Model assessment
Model identification
Model persistence and prediction
Machine learning models

Reinforcement learning techniques

Reinforcement learning training
Reinforcement learning cross-validation
Reinforcement learning hyperparameter selection

The preliminary version of the toolbox is now finalized and the project is entering the implementation phase where available functions are being combined and organized into pipelines tailored to answer the specific needs of each business case. As the project progresses, the pipelines will be packaged and made accessible directly via the BIGG platform.

Fully integrated in the BIGG architecture, the AI toolbox is leveraging the BIGG harmonized data format allowing to simplify the necessary steps of data understanding and data preparation. Although the planned AI toolbox is expected to be delivered as a fully packaged solution for the industry, the core components of the AI toolbox will be fully accessible to enable every stakeholder to assemble the components to best fit its own use. The functions are developed in both R and Python for maximum compatibility with existing platforms.

The first results of the Ai toolbox implementation are expected by the end of the month.