Data Driven Modeling

Technical seminar series on the main concepts and methods in probability theory, statistics, optimization, linear algebra and machine learning theory toward a framework for data literacy 

CAAD, Institute of Technology in Architecture, ETH Zurich 2016

A short description of the course

Ubiquitous data streams, being emitted from different aspects of our daily life has opened up new concepts of scientific modeling such as data driven modeling that challenge the classical notions of domain expertise, rule based systems and theory based models of real world phenomena. While we see data streams and Big Data as a new resource for modeling and design, we think its applications become prominent if we look at data via a new set of mathematical and coding skills, which goes beyond geometric thinking.

Toward this goal, in this series of technical seminars we introduce and discuss some of these mathematical concepts (as shown bellow) in four interlinked categories of Probability Theory and Statistics, Optimization Theories, Linear Algebra and Machine Learning Theories. All the concepts and techniques will be presented using Python programming language and sample data sets, while the objective is to attract architectural researchers to challenge these wild technologies from the perspective of design.

  • The codes and the presentations can be viewed here and downloaded from this Github repository (being updated every week).
  • All the codes are in Python.