Unique ID: 2017003
Division: | eScience Institute |
---|---|
Issue Date: | February 13th 2019 |
Last modified: | February 22nd 2019 |
Crowd-sensing Census
A Tool for Estimating and Visualizing Poverty Maps.
SDG: 01 - No Poverty
As increasingly more people seek to live in urban cities, governments and other organizations face the challenge of effectively identifying areas in most need of revitalization and intervention.
One way of designing such interventions is by using “poverty maps”.
Poverty maps are designed to simultaneously display the spatial distribution of welfare and different dimensions of poverty determinants. The plotting of such information on maps however heavily relies on data that is collected through infrequent national household surveys and censuses. However, due to the high cost associated with this type of data collection process, poverty maps are often inaccurate in capturing the current deprivation status.
In this project, we address this challenge by means of a methodology that relies on alternative data sources from which to derive up-to-date poverty indicators, at a very fine level of spatial granularity. We validate our methodology for the city of Milano. Based on our methodology and design requirements gathered from stakeholders we design and implement a poverty mapping tool for policy makers.
Project Objective:
Exploration, Other, Scientific / research
Project Outcomes:
To build a a tool that can estimate and visualise poverty maps for any region in the world. The tool allows the user to upload a CDR file, and in case of no CDR it relies on prediction based on OSM only (using API).
Publications Comments:
The current work is under submission.
Statistical Area
Project Sources
Type Of Institution: | research institute |
---|---|
Big Data Source: | Mobile phone data, OpenStreetMaps, iSTAT Census data |
Region: | Europe & Central Asia |
Country Area: | Italy |
Id Country Regional: | country |
Partnerships
Data Providers: | Mobile Phone operator |
---|---|
Other Partners: | Research or academic institute |
Partnership Comments: | This project is conducted as part of DSSG (Data Science for Social Good) in University of Washington. |
SDG Indicators
SDG: | 01 - No Poverty |
---|---|
SDG Comments: | This project aims to predict poverty based on big data thus eliminating the need for costly household surveys and census. |
Accessing Data
Data Access Rights: | Broader access rights |
---|---|
Data Access Comments: | Data made publicly available by Telecom Italia as part of their Data Challenge |
Data Coverage
Data Coverage: | Only a portion of all data |
---|---|
Coverage Geo Pop: | Part of country / high % of market |
Cost Implication: | Free |
Coverage Geo Comments: | Aggregated calls based on Cell Towers rather than individual's records |
Coverage Period: | 2 months |
Data Quality
Quality Aspects Evaluated: | Validity, Accuracy, including selectivity |
---|---|
Validation Comments: | Ground truth data about poverty collected from last Italian census available from iStat |
Quality Framework Comments: | Both quantitive and qualitative. Quantitive evaluation of the prediction model and qualitative model of the poverty maps. |
Data Quality Concerns Comments: | As iStat does not provide a single composite index of poverty we needed to construct this value based on multiple variables such as unemployment, education etc. by using Principle Component Analysis. |
Methodology
Methods Used: | Traditional statistical methods, Supervised learning, Data visualization methods, Machine learning (Random forest, etc.) |
---|
Technologies
Technologies: | GIS, Data visualization tools, Other, Hadoop Clusters, Cloud services |
---|---|
Technologies Comments: | R, Python |
Other
Income Level: | High-income |
---|---|
Iso: | IT |