Martin RΓΆck

Research Fellow @KU Leuven; Environmental Life Cycle Modelling of Buildings
Hi there! I am a built environment researcher focused on environmental modelling of buildings and building stocks using life cycle-based assessment methods

AI for life cycle emissions of buildings

Hi there!
Let me get this section started with a proposal for collaboration:
We have got the data, do you have the AI/ML? πŸ‘ΎπŸ˜œ


I am conducting research on the environmental modelling of buildings using life cycle assessment (LCA) for assessing impacts across their full life cycle. I.e. from material production over transport and construction, operation, maintenance and replacement, deconstruction and end-of-life treatment - you get the idea. As you may or may not know, building construction and operation contributes ~40% of global GHG emissions. So the building sector is obviously a major contributor to climate change, but also an area with strong potential for mitigation as well as need for adaptation.Β 


Over the past months/years we have been compiling a dataset on several hundreds of buildings and their emissions in different life cycle stages.
See this recent journal paper, where we present some high-level results from the analysis of the first batch of cases. Therein we showed that for effective climate change mitigation in the building sector it is crucial to address and reduce so-called embodied emissions, i.e. those related to construction material production and processing. Since that paper, we've advanced considerably and are at the point where we could use some ML support...

We now have detailed data on around 1.500 building case studies, including, e.g., information on building use, geometry, materials, assessment methods, ... and, most of all, the GHG emissions across the life cycle, in different life cycle stages, and for different building parts.

AI/ML application

There are two use cases for AI/ML which I have been exploring and would like to elaborate with the support of skilled ML researchers, hopefully to be found in this community:
1) Understanding the drivers of embodied/life cycle emissions (building design features,Β  contextual parameters, assessment methodology aspects), their influence and respective importance for both embodied and operational emissions
2) Training an ML model to predict embodied/life cycle emissions of buildings, based on basic information about building type, geometry, materials, etc.

Is it you?

To advance the research on this and get our analysis fit for publication, I am looking for a motivated AI/ML expert with
  • Experience in applying ML to patchy datasets with numeric and categorical values
  • Experience in applying decision trees + visualization, Shapley values for explaining feature influence, and/or other explainable ML approaches
  • Interest to work on this pro bono (no project funding currently) with the ambition to get it fit for a joint publication in a high-impact journal

If this sounds like you just get in touch and we take it from there!
Also, if you have questions on the above, just hook me up.
Looking forward to hear from you!


Keywords: regression/classification, categorical data, decision tree, shapley values
Like Comment