How Technology Can Be Used to Predict Poverty

POVERTY. The biggest world issue and the first Global Goal that the United Nations want to eradicate. In despite of the efforts made in the last 200 year around 836 million people still live in poverty and majority of them live on less that $1.25 a day what is defined by the World Bank as extreme poverty. By 2030 the goal is zero poverty and the fist step to eliminate it is to identify the regions most by plagued it. To know the poverty level is necessary to collect data and information and there is currently almost no information about it because of the lack of resources and efforts to avoid the problem in areas especially affected as Asia or Africa mainly.

According to a new study released recently in the journal Science, Scientists of Stanford University have found an accurate method, which allows coping with lacking of information problems. It combines machine learning algorithms and satellite imagery, which use the night-lights and high-resolution imagery to identify poverty areas. Initially a satellite collects night-lights data and daytime images that will be added to economic survey data in order to reach an accurate reality. Then the model takes all the information and uses the map to make predictions that fills gaps where there isn’t good data. By looking for correlation the model can associate areas with houses and cars are likely to be more lit or areas with more cars are more likely to have higher household spending in surveys.

So far, researchers have been using personal survey as the common method to collect data about poverty. They use to send surveys to every house with many questions about income and consumption and results are used to construct poverty measures. However, this is not an accurate method and it is typically expensive, time-consuming and sometimes impossible to conduct because, for instance, some countries from Africa as Nigeria, Tanzania, Uganda, Rwanda and Malawi live in permanent armed conflicts. The experiment was first carried out in these countries and results were surprisingly predictive of economic livelihoods. In order to check out, imagery from countries for which survey data was available were used to validate the computer model’s finding.

Although at the moment the available data only allows to know the currently poverty state, European Space Agency and other world organizations in order to help out this study expect to provide new satellite imagery in the near future to keep them collecting information across the world and making future predictions.

The innovative system only works in the 5 countries where is being tested, but all efforts are being made to have it working across the world as soon as possible because of the high benefits to predict and eradicate poverty. One more time it is showed how to invest in technology innovations can impact positively in people lives.

To learn more about combining satellite imagery and machine learning to predict poverty: