Credit: both the logo and graphics above is created by ChatGPT with the author's instruction

• MUSA 5000 Spatial Statistics (with Haoyu Zhu & Kavana Raju)
• MUSA 5080 Public Policy Analytics More will come soon.

The Master of Urban Spatial Analytics program is training a new generation of data scientists to tackle complex public policy problems. Using geo-spatial computing methods and open source software tools, our students and faculty seek to create positive change through data-driven decision-making. Responsible, domain-savvy data scientists can enable governments to understand what works and decide how to deploy limited resources to benefit the public. In the MUSA program, we aren’t training engineers; we are empowering social science students to use technology to solve problems that they find meaningful.

Instructor: Prof. Eugene Brusilovskiy, Lectuer, Weitzman School of Design

This hands-on course will cover a wide range of methods frequently used for analyzing urban and spatial data. These methods are drawn from a variety of fields, including traditional statistics, spatial econometrics, and machine learning

1. Regression analysis (OLS, ridge/lasso, logistic, multinomial logit);
2. Measures of spatial autocorrelation: (spatial lag, spatial error regression, and Geographically weighted regression (GWR)
3. Spatial regression (spatial lag, spatial error, geographically weighted regression);
4. Point pattern analysis;
5. An introduction to clustering methods (k-means, hierarchical clustering, DBSCAN);
6. Big data and GIS.

Students will learn the assumptions and limitations of each method, and assignments will focus on the implementation, presentation, and interpretation of the analyses. Students will use R and GeoDa in this course.

Instructor: Dr. Elizabeth Delmelle, Associate Professor, Director of MUSA, Weitzman School of Design

This course teaches advanced spatial analysis and an introduction to data science/machine learning in the urban planning and public policy realm. The class focuses on real-world spatial analysis applications and, in combination with introductory machine learning, provides students a modern framework for efficiently allocate limited resources across space. Unlike its private sector counterpart, data science in the public or non-profit sector isn't strictly about optimization - it requires understanding of public goods, governance, and issues of equity. We explore use cases in transportation, housing, public health, land use, criminal justice, and other domains. We will learn novel approaches for understanding and avoiding risks of "algorithmic bias" against communities/people of color as well as communities of different income levels.

The format of the class includes weekly lectures/in-class demos and labs. There are seven required assignments, including two projects. The class is conducted entirely in R. Having experience in R and the ‘tidyverse’ is helpful but not strictly required.