This research reveals how credit card transactions combined with mobile phone data can uncover distinct urban lifestyles, with applications ranging from policy design to gender-disaggregated economic analysis.

Credit card transactions combined with mobile phone data can be analyzed to uncover distinct types of spending habits and lifestyles at unprecedented scale. These lifestyles present consistent relationships with age, total expenditure, gender, and the diversity in mobility and social networks of identified individuals.

Just as motifs in network science provided important insights into networks with power-law degree distributions, the lifestyles extracted from labeled spending data reveal essential information about human behavior. This project has a twofold aim: first, discovering the existence of ubiquitous trends in spending behavior patterns and how these trends connect with other attributes of our social life; second, developing a new tool for big data analysis to overcome several limitations of traditional machine learning algorithms.

We have developed a novel framework to analyze credit card transaction records that enables the identification of characteristic signatures of human lifestyles. Our method, for the first time in the academic literature, couples credit card records with mobile phone data and demographic characteristics of individual users to provide a comprehensive picture of human behavior at unprecedented scale.

This analysis unveils relationships between human mobility, spending behavior, and social activities. We identified six distinct urban lifestyle clusters: Commuters, Homemakers, Young, Hi-Tech, Average, and Dinner-out. Each lifestyle shows distinctive patterns in terms of spending categories, mobility patterns, and social network diversity.

Over a longer timeframe, such analysis could reveal signals about how people cope with a wide range of environmental and economic shocks. Understanding changes in expenditure sequences of vulnerable population groups under external shocks—such as natural disasters or economic adversities—will be useful for developing early warning and monitoring mechanisms to inform public policies that protect these populations.

How can we predict someone's shopping behavior from their mobility patterns? This book chapter tackles this challenge by jointly modeling the lifestyles of individuals through collective matrix factorization—viewing shopping and mobility as two aspects of the same underlying lifestyle.

Using latent Dirichlet allocation, we first identify five distinct shopping behaviors from credit card sequences: food-centered purchases, business-related spending, luxury consumption (cable, department stores, air travel), tech-savvy patterns (computer services, gas stations), and subscription-based expenditure. We then transform cellular tower location data using Points of Interest categories crawled from Google's API, creating tower "classes" defined by nearby establishments like hospitals, universities, cafes, or car repair shops.

The key innovation is connecting these two views through collective matrix factorization, which assumes that shopping patterns and mobility patterns are generated from the same latent lifestyle information. This approach achieves a 1.3% reduction in prediction error compared to shopping-only models—demonstrating that where people go reveals information about what they buy. The recovered lifestyles show intuitive dual patterns: for instance, wealthier "urban white collar" lifestyles combine luxury shopping with high mobility diversity and visits to university and business areas, while food-oriented lifestyles show visits to cafes, gyms, and convenience stores.

Conventional data sources—household surveys, national accounts, institutional records—struggle to capture detailed information on the economic lives of women and girls. In this UN Foundation report, we demonstrate how credit card and mobile phone data can help close the global gender data gap by analyzing patterns of women's expenditure and mobility in a major Latin American metropolis.

Using over 10 weeks of anonymized transactions from 150,000 users, we find that food-related purchases dominate women's spending, with over a quarter of transactions in grocery stores, restaurants, and food shops. However, striking differences emerge between sexes: women have more transactions in grocery stores, insurance, and department stores, while men dominate restaurants and transport. Women also report less total expenditure per capita, indicating either reduced access to economic resources or different credit card usage patterns.

By combining credit card sequences with cell phone mobility data—including metrics like radius of gyration, mobility diversity, and social network diversity—we identified seven distinct economic lifestyle clusters among women. These "portraits of economic lifestyles" illustrate women's needs and priorities in ways that traditional surveys cannot capture. Over longer timeframes, such analysis could reveal real-time signals about how women cope with economic recessions, natural disasters, and other shocks—critical information for designing gender-responsive policies.