Mobility Modelling

Lecture 12 - Mobility as Characteristic for Socio-economic Analysis

8 May 2023

Mozhgan Pourmoradnasseri, Ph.D.

Is snow-clearing sexist?

Perez, Caroline Criado. Invisible women: Data bias in a world designed for men. Abrams, 2019.

Urban planning is gender-blind, disaggregated data is limited.
In several countries, 2/3 of public transport passengers are women.
Men, on average, have simple travel patterns: twice daily commute.
Women are more likely to trip chain.

Data on pedestrian injuries in Sweden showed 79% of accidents happened in winter.
Of them, 69% were women.
2/3 of injured had slipped on icy surfaces.
Only in one county (Skane), the estimated cost of healthcare for these injuries was 36 million Korons.
This is at least twice the cost of road winter maintenance.

Mobility does not occur in a neutral space with no significance.

Most studies on urban commuting focus on commuting between home and work; however, these trips constitute only 30% of daily trips.
Travel to work is different in nature. It happens in rush hours. People tolerate longer distances for going to work.
To have an inclusive vision, all forms of daily trips have to be considered.
Several studies indicate different commuting patterns between men and women.
Women visit fewer unique locations than men and distribute their time, less equally between them.

Gauvin, Laetitia, et al. "Gender gaps in urban mobility." Humanities and Social Sciences Communications 7.1 (2020): 1-13.

Gender gap in urban mobility

Data:

Mobile phone data of more than 400k unique users, with at least one call per day, for the duration May, June, July 2016 for the city of Santiago, Chile.
Gender and income information of users.
Census data, including employment gender ratio, education, size of household, ...
GTFS data, locations of public transport stations.
Points of interest from OSM.

Mobility Metrics:

Gauvin, Laetitia, et al. "Gender gaps in urban mobility." Humanities and Social Sciences Communications 7.1 (2020): 1-13.

Gender gap in urban mobility

\(N_l\); number of distinct locations visited by a user
\(\hat{N}_l\); the set of locations that account for 80% of a user’s calling activity.
Shannon entropy of user’s trajectories, where \(L\) is the full set of locations visited by a user, and \(p_l\) is the probability of observing a user in \(l\).
Radius of gyration of each user, \(r_g\) defined as the root-mean-square distance of an individual's movements from their centroid.

Gauvin, Laetitia, et al. "Gender gaps in urban mobility." Humanities and Social Sciences Communications 7.1 (2020): 1-13.

Gender gap in urban mobility

a. Violin plots show the distributions by gender of the number of locations accounting for 80% of a user’s activity.

b. the users’ Shannon mobility entropy.

c. Women visit fewer locations and their movements are characterized by a smaller entropy. Panel c shows the distributions of the mean probability of visiting the 5 most frequented locations of each user, by gender.

Gauvin, Laetitia, et al. "Gender gaps in urban mobility." Humanities and Social Sciences Communications 7.1 (2020): 1-13.

Gender gap in urban mobility

Gauvin, Laetitia, et al. "Gender gaps in urban mobility." Humanities and Social Sciences Communications 7.1 (2020): 1-13.

Gender gap in urban mobility

What are the strengths and weaknesses of this paper?
Can you think of other metrics?
How could this work be improved?
What data types can be used to analyze socioeconomic/demographic gaps?

What are your thoughts about this paper

Mobility Patterns and income segregation

Moro, Esteban, et al. "Mobility patterns are associated with experienced income segregation in large US cities." Nature communications 12.1 (2021): 4633.

Data:

4.5 million mobile phone users for six months.
a collection of ~1.1 million verified venues obtained via the Foursquare API in 11 large American cities.
Census data, income, population, ...

Method:

Home is estimated as the most frequently visited location during the night.
Income is estimated as the median of income in the area of residence.
Extract any visits an individual makes to a given place that lasts for more than 5 min.
To measure the income segregation of each place α in the city, the proportion of total time spent at that place by each income quartile is computed.

Mobility Patterns and income segregation

a. People from different census block groups visit a given place. Using the median income of each census block group, we calculate the distribution of time spent by the four income groups (quartiles) in that place. Income segregation of the place measures the unevenness of this distribution.

b. For a given individual, we calculate the distribution of time that individual encounters people of different income groups in each place and in total.

c. Map of the places in downtown Boston color-coded according to their income segregation.

Mobility Patterns and income segregation

Average place segregation by category as a function of the average distance traveled by individuals to reach that place from their home. Colors correspond to different groups of place categories and the size is proportional to the number of places in each category. As we can see the average income segregation depends both on the type of place and distance traveled.

What are your thoughts about this paper

Candipan, Jennifer, et al. "From residence to movement: The nature of racial segregation in everyday urban mobility." Urban Studies 58.15 (2021): 3095-3117.

Racial segregation in Us cities

Data:

133,766,610 geotagged tweets sent by over 375,000 Twitter users in the 50 largest US cities, from 1 October 2013 to 31 March 2015.
Racial composition of neighborhoods; majority Black, majority White, majority Hispanic and mixed race (no majority)

Metric:

Defined segregated mobility index (SMI): average proportion of visits by residents from a block group to all other block groups.

Results:

Confirms previous findings.

residential segregation --> impacts life opportunities

Daily mobility:

Could reduce negative effects of segregation if people move and mix.
Could deepen segregation by limitations in accessibility.

Bierbaum, Ariel H., Alex Karner, and Jesus M. Barajas. "Toward mobility justice: Linking transportation and education equity in the context of school choice." Journal of the American Planning Association 87.2 (2021): 197-210.

Daily Mobility Patterns:

Reducing or Reproducing Inequalities and Segregation?

Hedman, Lina, et al. "Daily mobility patterns: Reducing or reproducing inequalities and segregation?." Social Inclusion 9.2 (2021): 208-221.

Data:

The population register database.
Mobile data of Thursday, 28 March, and Saturday, March 30, 2019, with market share of 10–20%.

Method:

Home location is estimated similar to previous works.
For day activities, midpoints are calculated using the weighted coordinates of the phone between 10 am and 12 am,1 pm and 3 pm.

Daily Mobility Patterns:

Reducing or Reproducing Inequalities and Segregation?

Hedman, Lina, et al. "Daily mobility patterns: Reducing or reproducing inequalities and segregation?." Social Inclusion 9.2 (2021): 208-221.

Falun-Borlänge, with 7.6% of the nonEuropean population.
Gävle-Sandviken, with 10.9% of the nonEuropean population.
category 1, areas with a low share (below the mean, decile 1–7); category 2, areas with a relatively high share (around or above the mean, deciles 8–9); and category 3, areas with a high share of nonEuropeans.

Daily Mobility Patterns:

Reducing or Reproducing Inequalities and Segregation?

Hedman, Lina, et al. "Daily mobility patterns: Reducing or reproducing inequalities and segregation?." Social Inclusion 9.2 (2021): 208-221.

Daily Mobility Patterns:

Reducing or Reproducing Inequalities and Segregation?

Are younger age groups less segregated?

Silm, Siiri, Rein Ahas, and Veronika Mooses. "Are younger age groups less segregated? Measuring ethnic segregation in activity spaces using mobile phone data." Journal of Ethnic and Migration Studies 44.11 (2018): 1797-1817.

Data:

Enriched mobile data; including gender, year of birth, and preferred communication language (E, R, E), for 2011.
Users are randomly selected according to the population distribution of 2011 data (age/gender/ethnicity).

Metrics:

Index of dissimilarity (ID); a measure of (un)evenness, showing the degree to which two populations are distributed differently. 0 indicates a completely even distribution (integration), and 1 indicates a completely uneven distribution (segregation).
Entropy to characterises the heterogeneity of the visitation patterns.

Are younger age groups less segregated?

Metrics:

Index of dissimilarity (ID); a measure of (un)evenness, showing the degree to which two populations are distributed differently. 0 indicates a completely even distribution (integration), and 1 indicates a completely uneven distribution (segregation).
Calculated for every age group and residence, workplace, and out-of-home non-employment district; where \(e_i (r_i)\) is the number of Estonian (Russian) -speakers in study district \(i\); \( E (R)\) is the total number of Estonian (Russian) -speakers in all study districts.