Public Policy Analytics

Code and Context for Data Science in Government

Ken Steif

• Publisher: CRC Press
• Year: 2021
• Print ISBN:
  9780367507619
  9780367516253
• Edition: 1
• eBook ISBN:
  9781000401615
• Topic: Politics & International Relations
• Subtopic: Statistics for Business & Economics
• Index: Politics & International Relations

Public Policy Analytics: Code & Context for Data Science in Government teaches readers how to address complex public policy problems with data and analytics using reproducible methods in R. Each of the eight chapters provides a detailed case study, showing readers: how to develop exploratory indicators; understand 'spatial process' and develop spatial analytics; how to develop 'useful' predictive analytics; how to convey these outputs to non-technical decision-makers through the medium of data visualization; and why, ultimately, data science and 'Planning' are one and the same. A graduate-level introduction to data science, this book will appeal to researchers and data scientists at the intersection of data analytics and public policy, as well as readers who wish to understand how algorithms will affect the future of government.

1

Indicators for Transit-oriented Development

DOI: 10.1201/9781003054658-1

1.1 Why start with indicators?

According to the Federal Transit Administration (FTA), not one of America's largest passenger subway systems saw fare revenues exceed operating expenses in 2015.¹

This is an indicator - a stylized fact that gives simple insight into a complicated phenomena. Mastering indicators is critical for conveying nuanced context to non-technical audiences. Here are four suggestions on what makes a good indicator:

1. A relatable indicator is typically motivated by a pressing policy concern. “How is it possible that passenger rail in New York City has such widespread delays, service suspensions, and rider discontent?” A great indicator solicits interest from an audience.
2. A simple indicator may be used as an exploratory tool in place of more complex statistics. Simplicity helps the audience understand the indicator's significance and keeps them engaged in the analysis.
3. A relative indicator draws a contrast. “How can New York City passenger rail, with the most trips, still lose more money than each of the next ten largest cities?” Contextualizing an indicator with a relevant comparison makes for greater impact.
4. A good indicator typically generates more questions than answers. Thus, a good indicator fits into a broader narrative, which helps motivate a more robust research agenda and ultimately, more applied analytics.

¹ Federal Transit Administation. “NTD Data, 2015”. https://www.transit.dot.gov/ntd/ntd-data

Simplicity is an indicator's strength, but it may also be its weakness. Most statistics make assumptions. You should be aware of these assumptions, how they affect your conclusions, and ultimately how the audience interprets your results.

In this first chapter, space/time indicators are built from U.S. Census data to explore transit-oriented development (TOD) potential in Philadelphia. Along the way, we will learn how assumptions can lead to incorrect policy conclusions.

TOD advocates for increased housing and amenity density around transit (rail, subway, bus, etc.). There are many benefits to promoting this density, but two examples are particularly noteworthy.

First, transit needs scale to exist. Transit demand is a function of density, and the more households, customers, and businesses around transit, the more efficient it is to operate a transit system. Efficiency means less maintenance, staffing, etc. Interestingly, Figure 1.1 suggests that most transit systems are remarkably inefficient despite being in cities with density of just about everything.

FIGURE 1.1 Revenue per passenger for passenger subway systems in the United States.

Second, TOD is important for land value capitalization, which is essential to both developers and governments. If renters and home buyers are willing to pay more to locate near transit amenities, it should be reflected in higher land values and property tax returns near stations.

In this chapter, we play the role of Transportation Planner for the City of Philadelphia and assess whether rents are higher in transit-rich areas relative to places without transit access. If residents value these locations, officials might consider changing the zoning code to allow increased density around transit.

As the analysis progresses, spatial data wrangling and visualization fundamentals are presented with the tidyverse, sf, and ggplot2 packages. The tidycensus package is used to gather U.S. Census tract data. We begin by identifying some of the key assumptions made when working with geospatial census data.

1.1.1 Mapping and scale bias in areal aggregate data

Data visualization is a data scientist's strongest communication tool because a picture tells a thousand words. However, visualizations and maps in partic...

Table of contents

• Cover
• Half Title
• Series Page
• Title Page
• Copyright Page
• Contents
• About the Author
• Preface
• Introduction
• 1 Indicators for Transit-oriented Development
• 2 Expanding the Urban Growth Boundary
• 3 Intro to Geospatial Machine Learning, Part 1
• 4 Intro to Geospatial Machine Learning, Part 2
• 5 Geospatial Risk Modeling - Predictive Policing
• 6 People-based ML Models
• 7 People-based ML Models: Algorithmic Fairness
• 8 Predicting Rideshare Demand
• Conclusion - Algorithmic Governance
• Index