A platform for modeling metropolitan real estate markets
This new code base is a streamlined complete re-implementation of the longstanding UrbanSim project aimed at reducing the complexity of using the UrbanSim methodology. Redesigned from the ground up, the new library is trivial to install, the development process is made transparent via this GitHub site, and exhaustive documentation has been created in the hopes of making modeling much more widely accessible to planners and new modelers.
We lean heavily on the PyData community to make our work easier - Pandas, IPython, and statsmodels are ubiquitous in this work. These Python libraries essentially replace the UrbanSim Dataset class, tools to read and write from other storage, and some of the statistical estimation previously implemented by UrbanSim.
This makes our task easier as we can focus on urban modeling and leave the infrastructure to the wider Python community. The Pandas library is the core of the new UrbanSim, which is an extremely popular data manipulation library with a large community providing support and a very helpful book.
We have now converted a full set of UrbanSim models to the new framework, and have running applications for the Paris, Albuquerque, Denver, Bay Area, and Detroit regions. We have implemented a complete set of hedonic price models, location choice models, relocation and transition models, as well as a new real estate development model using proforma analysis.
We do strongly recommend that you contact the team at www.urbansim.com about your project to make sure you can get support when you need it, and know what you are getting into. For major applied projects, professional support is highly recommended.
