Interpretable Poverty Mapping using Social Media Data, Satellite Images, and Geospatial Information

Ledesma, Chiara; Garonita, Oshean Lee; Flores, Lorenzo Jaime; Tingzon, Isabelle; Dalisay, Danielle

doi:10.48550/arxiv.2011.13563

Cited by 7 publications

(10 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The methodological differences are presented in four aspects: the data source, the feature extraction methods, the prediction methods, and the target variable chosen. The models used the census or survey data geolocalized, which contains the ground truth variable; the primary source in several studies is the DHS survey that meets the two requirements: the household's geolocation and the wealth index as a proxy of social condition (Blumenstock et al 2015;Ledesma et al 2020;Lee and Braithwaite 2020;Sheehan et al 2019;Steele et al 2017;Weidmann and Schutte 2017). Some works focus on census data as the main source (Engstrom et al 2017;Pandey et al 2018;Pokhriyal and Jacques 2017;Pokhriyal et al 2020), while others focus on local or national surveys (Ayush et al 2020;Gebru et al 2017;Steele et al 2017;Watmough et al 2019).…”

Section: Related Workmentioning

confidence: 99%

“…Some works focus on census data as the main source (Engstrom et al 2017;Pandey et al 2018;Pokhriyal and Jacques 2017;Pokhriyal et al 2020), while others focus on local or national surveys (Ayush et al 2020;Gebru et al 2017;Steele et al 2017;Watmough et al 2019). In addition to using satellite imagery, some studies combine data from different sources, such as Wikipedia geolocated articles (Sheehan et al 2019), the settlement data from the United Nations (Lee and Braithwaite 2020), the points of interest from OpenStreetMaps (Hu et al 2022;Ledesma et al 2020;Lee and Braithwaite 2020), Google street view images (Gebru et al 2017), counting of users from Facebook (Ledesma et al 2020), aerial images (Pokhriyal et al 2020), indicators from open platforms Niu et al (2020), street level images Suel et al (2021), and call detail records (Blumenstock et al 2015;Moya-Gómez et al 2021;Pokhriyal and Jacques 2017;Pokhriyal et al 2020;Steele et al 2017). The combination of data sources has demonstrated promising results for multidimensional poverty estimation (Pokhriyal and Jacques 2017;Pokhriyal et al 2020).…”

Section: Related Workmentioning

confidence: 99%

“…Furthermore, a multitask fully connected neural network has been implemented for predicting the roof type, source of lighting, and the source of drinking water for rural villages in India (Pandey et al 2018), and in a simple fully connected network for predicting the international wealth index (Sheehan et al 2019). In addition, there are works using extraction methods such as geospatial environmental models (Watmough et al 2019), counting methods based on descriptive statistics or linear transformations (Ledesma et al 2020;Niu et al 2020;Weidmann and Schutte 2017), gaussian process regression (Pokhriyal et al 2020), document representation (Sheehan et al 2019), and non-spatial generalized linear models (Steele et al 2017).…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Predicting Multidimensional Poverty with Machine Learning Algorithms: An Open Data Source Approach Using Spatial Data

Muñetón-Santa,

Manrique-Ruiz

2023

Social Sciences

View full text Add to dashboard Cite

This paper presents a methodology to estimate the multidimensional poverty index using spatial data at the street block level. The data used in this study were obtained from Open Street Maps and ESA’s land use cover, which are freely available sources of spatial information. The study employs five machine-learning algorithms, including Catboost, Lightboost, and Random Forest, to estimate the multidimensional poverty index with spatial granularity. The results indicate that these models achieve promising performance in predicting poverty levels in Medellín, Colombia. The results showed that the Random Forest algorithm achieved the highest performance, with an MAE of 0.07504. Furthermore, the spatial distribution of the multidimensional poverty estimate was highly correlated with the true values of the distribution. This work contributes to predicting multidimensional poverty by demonstrating the potential of machine learning algorithms to utilize accessible spatial data. By providing evidence of the feasibility of estimating poverty levels at a granular spatial level, this methodology offers a powerful tool for policymakers to make poverty social interventions with low-cost evidence. Furthermore, this study has important implications for poverty eradication efforts in developing countries, where access to reliable data remains challenging.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Predicting Multidimensional Poverty with Machine Learning Algorithms: An Open Data Source Approach Using Spatial Data

Muñetón-Santa,

Manrique-Ruiz

2023

Social Sciences

View full text Add to dashboard Cite

show abstract

“…More recently, machine learning methods have been used to estimate asset wealth and poverty across multiple countries using non-survey datasets such as satellite imagery [7,9,10,15,20,21,23], call data records [17,19], social media activity [12] or a combination of these [3]. Notably, these works show that greater than 50% of the variation in survey-measured poverty can be explained with satellite image-based poverty predictions [5].…”

Section: Introductionmentioning

confidence: 99%

Poverty rate prediction using multi-modal survey and earth observation data

Fobi

Cardona²,

Collins

et al. 2023

Proceedings of the 6th ACM SIGCAS/SIGCHI Conference on Computing and Sustainable Societies

View full text Add to dashboard Cite

This work presents an approach for combining household demographic and living standards survey questions with features derived from satellite imagery to predict the poverty rate of a region. Our approach utilizes visual features obtained from a single-step featurization method applied to freely available 10m/px Sentinel-2 surface reflectance satellite imagery. These visual features are combined with ten survey questions in a proxy means test (PMT) to estimate whether a household is below the poverty line. We show that the inclusion of visual features reduces the mean error in poverty rate estimates from 4.09% to 3.88% over a nationally representative outof-sample test set. In addition to including satellite imagery features in proxy means tests, we propose an approach for selecting a subset of survey questions that are complementary to the visual features extracted from satellite imagery. Specifically, we design a survey variable selection approach guided by the full survey and image features and use the approach to determine the most relevant set of small survey questions to include in a PMT. We validate the choice of small survey questions in a downstream task of predicting the poverty rate using the small set of questions. This approach results in the best performance -errors in poverty rate decrease from 4.09% to 3.71%. We show that extracted visual features encode geographic and urbanization differences between regions.

show abstract

“…In crowdsensing, interested participants connected through crowdsensing apps or websites are engaged in carrying out specialized distributed sensing tasks through different crowdsensing platforms (e.g., mobile apps such as Citizen and Waze). Examples of social sensing applications include studying human mobility in urban areas [9]; obtaining situation awareness in the aftermath of disasters [10], poverty prediction and mapping [11], locating power outages in cities [12], urban land usage classification [13], and contact tracing of contagious diseases such as COVID-19 [14]. Figure 1 (a) and (b) illustrate examples of physical and social sensing applications respectively.…”

Section: Introductionmentioning

confidence: 99%

A Survey on Social-Physical Sensing: An Emerging Sensing Paradigm that Explores the Collective Intelligence of Humans and Machine

Rashid¹,

Wei²,

Wang³

2021

Preprint

View full text Add to dashboard Cite

Propelled by the omnipresence of versatile data capture, communication, and computing technologies, physical sensing has revolutionized the avenue for spontaneously capturing and interpreting the real-world phenomenon. Despite its virtues, various limitations (e.g., high application specificity, partial autonomy, and sparse coverage) hinder physical sensing's effectiveness in critical scenarios such as disaster response and urban anomaly detection. Meanwhile, social sensing is contriving as a pervasive sensing paradigm that leverages the observations from human participants equipped with portable devices and ubiquitous Internet connectivity (i.e., through social media or crowdsensing apps) to perceive the occurrences in the environment. While social sensing possesses a plethora of benefits, it also inherently suffers from a few drawbacks (e.g., inconsistent reliability, uncertain data provenance, and limited sensing availability). Motivated by the complementary virtues of both physical and social sensing, social-physical sensing (SPS) is protruding as an emerging sensing paradigm that tightly integrates the social and physical sensors at an unprecedented scale. The vision of SPS centers on mitigating the individual weaknesses of physical and social sensing while exploiting their collective strengths in reconstructing the "state of the world", both physically and socially. While a good amount of interesting SPS applications has been explored, several important unsolved challenges and open research questions prevail in the way of developing innovative and practical SPS systems, which require careful scrutiny and research to address. In this paper, we provide a comprehensive survey of SPS, with an emphasis on its definition and key enablers, state-of-the-art applications, potential research challenges, and the road-map for future work. While existing survey papers related to sensing have discussed sensing schemes that are partly related to SPS, they do not explicitly provide an extensive overview of the SPS paradigm or the state-of-the-art SPS schemes. This paper intends to bridge this knowledge gap by thoroughly examining the various aspects of SPS crucial for building potent SPS systems.

show abstract

Interpretable Poverty Mapping using Social Media Data, Satellite Images, and Geospatial Information

Cited by 7 publications

References 0 publications

Predicting Multidimensional Poverty with Machine Learning Algorithms: An Open Data Source Approach Using Spatial Data

Predicting Multidimensional Poverty with Machine Learning Algorithms: An Open Data Source Approach Using Spatial Data

Poverty rate prediction using multi-modal survey and earth observation data

A Survey on Social-Physical Sensing: An Emerging Sensing Paradigm that Explores the Collective Intelligence of Humans and Machine

Contact Info

Product

Resources

About