Easy peasy massive parallel computing in R (Mikkel)
Mikkel a member and Co-Organizer of CopenhagenR paid a short visit to Vienna and before heading to Ennio Moricones (Vienna) concert he showed us around on the functionality of the ‘future’ and ‘furrr’ R packages.
These provide a framework that make it possible to write code, that works seamlessly on a laptop or on a supercomputer.
His R-talk showed us how.
We ran through a concrete example. You can find a fork of his docker project at ViennaR’s github repo easy_peasy.
Evaluating Austrias social housing program (Resul)
In his presentation, he showed example of applied microeconomics (utility functions) using modern statistical methods (deep learning models), to evaluate Austrian social housing programs (gemeindewohnung and genossenschaftswohnung).
Estimating the equivalent-variation measures for 282 Austrian households living in social housing using EU-SILC data for the year 2008.
Household preferences were represented by a Cobb-Douglas utility function whose parametrization was based on the rent-to-income ratios. The market values of the subsidized housing units were obtained using a hedonic regression which he estimated by a deep learning model, based on attributes of the units contained in the data. This was one of the first times that Austrian social housing got evaluated this way. Data impoutation was done via mice algorithm (ie using the R package with the same name). The results are interesting for anyone living in a rented dwelling in Austria. The detailed presentation is available here.