Predicting Gentrification in Mexico City using Neural Networks

Palafox Novack, Leon Felipe; Ortiz-Monasterio, Pedro

doi:10.1109/IJCNN48605.2020.9207685

Predicting Gentrification in Mexico City using Neural Networks

Journal

2020 International Joint Conference on Neural Networks (IJCNN)

Date Issued

2020

Author(s)

Palafox Novack, Leon Felipe

Facultad de Ingeniería - CampCM

Ortiz-Monasterio, Pedro

Facultad de Ingeniería - CampCM

Type

text::conference output::conference proceedings::conference paper

DOI

10.1109/IJCNN48605.2020.9207685

URL

https://scripta.up.edu.mx/handle/20.500.12552/3979

Abstract

Gentrification is a process that affects millions of people every year. In this process, high-income residents replace low-income residents in neighborhoods near city centers and business centers. The gentrification process drives low-income people to find unfamiliar places to live, which in and of itself brings multiple social problems about housing, transportation, and schooling.Many studies in social geography have looked at the onset of gentrification in a neighborhood. Yet, studying gentrification in a single neighborhood is a slow process that requires expertise about the different elements that can cause it, like housing prices, businesses in the neighborhood, and other social elements. Each of these elements is idiosyncratic to each country and area.In this work, we mix the predictive power of Neural Networks with an Interpretability Method called LIME, which helps understand which factors are driving the classification given the data and a trained classifier. With this, we expect to have an overall model that gains a deeper understanding of which effects drive gentrification in different settings. © 2020 IEEE.