SeptembeR Meetup: R for Real Estate and Machine Learning Model Summaries

We are looking forward to the next Zurich R User Meetup, sponsored by Datahouse.

***
Schedule:
06.15 pm Doors open
06.30 pm Introduction / Welcome by the organizing team
06.40 pm Talks (see below)
ca 07.40 Take the Stage: Pitching job ads, ideas, events, ...
ca 07.45 - 09.00 pm Apéro
****

Valuation of Residential Real Estate using R
Simon Stehle, Wüest Partner AG

Transparency is a key element of well-functioning markets. Particularly in markets with heterogeneous and infrequently traded goods such as real estate, independent valuation is thus key to ensure efficient transactions - reducing frictions and risks. In this presentation, I show why and how simple tools like linear regression are used to setup models that are widely respected and used in the industry. While doing so, I shed light on theory, industry priorities, and further applications.
***

Concise, Comprehensive, and Customizable Summaries for Machine Learning Models
Susanne Dandl, Epidemiology, Biostatistics and Prevention Institute, Universität Zürich

In machine learning (ML), transparency and interpretability are key to building trust and supporting informed decisions. Interpretable ML, or Explainable AI, addresses these needs by developing techniques for post-hoc interpretations of trained models. In this talk, I introduce a novel R package focused on performance measures and interpretation methods. Inspired by the well-known summary() function for (generalized) linear models in R, the mlr3summary package provides a concise and informative summary of model performance, model complexity, as well as variable importances and effects, for non-parametric ML models. By making use of mlr3 – a rich R package ecosystem for applied ML - the package’s functionality hopefully enhances model transparency and comparability across a wide range of ML applications.