XClose

UCL Grand Challenges

Home
Menu

3-to-1: Enhancing Indicator Quality for Sustainable Development Goals Through Big Data

Transformative Technology funding awarded in 2019/20

Satiliette crossing the earth

1 October 2019

Grant


Grant: Grand Challenges Small Grants
Year awarded: 2019-20
Amount awarded: £7,433

Academics 


  • Aidan O’Sullivan, UCL Energy Institute
  • Paul Ekins, Institute for Sustainable Resources

Progress towards the UN Sustainable Development Goals is measured against a set of indicators for each goal. However across the numerous indicators associated with each target there are varying levels of quality of data availability. The indicators are ranked according to a Tier classification, with Tier 1 indicators being ones for which high quality data is available to assess progress. In total there are 93 Tier 1 indicators, 72 Tier 2 indicators and 62 Tier 3 indicators. Tier 3 indicators are characterised as "No internationally established methodology or standards are yet available for the indicator".

The objective for this project would be to take a subset of these Tier 3 indicators and using modern techniques of machine learning and data fusion combine disparate available sources of data to develop a Tier 1 level data set to enable the indicator to be measured. The first step in the project, would be to identify a subset of Tier 3 indicators for which suitable sources of data are available. This could be satellite imaging data, social media data, telecommunication data, text data from newspaper articles etc. Another goal, to take action to tackle climate change is of particular interest and a likely area of focus will be explored. With the indicators and relevant datasets identified a host of machine learning techniques from Neural Networks to Natural Language Processing will be deployed to identify data quality issues measure gaps and enhance the quality of the data to assess the SDG indicator against. Data fusion methods will be used to combine multiple sources of complementary data and probabilistic methods used to enhance data quality. Validation will involve benchmarking against known sources of high quality data for example; inequality measures for which data is available in the UK but not for less developed countries.

Outputs and Impacts