ITEM: A researcher at Telenor Group Research says he has developed a machine learning algorithm that uses mobile phone call records to determine literacy rates in developing markets by location.
According to MIT Technology Review, researcher Pål Sundsøy says he started with a regular survey – conducted by a professional agency for a mobile operator – covering 76,000 mobile users in Asia that collected the user’s mobile number and asked them if they could read.
Sundsøy matched that data against the mobile operator’s call data records, which enabled him to work out where each user was, who they called, for how long, etc. Then he crunched 75% of the correlated data to detect patterns with illiterate users, and used the remaining 25% to see if those patterns could identify illiterate people and areas where there is a higher proportion of illiterate people.
All in all, he says, his machine learning algorithm can spot illiterate individuals with surprising accuracy. “By deriving economic, social, and mobility features for each mobile user we predict individual illiteracy status with 70 percent accuracy,” he says, pointing out that this allows areas with low literacy rates to be mapped.
The algorithm still needs further testing in other locations, but if it works, it will be a boon for aid agencies tackling illiteracy who need a better idea of where to place their resources for doing so. The usual way to do this is household surveys, which are expensive, time consuming and thus hard to update frequently. Machine-learning algorithms coupled with mobile network data could be a way to make such research faster and cheaper.
Some 750 million people around the world are unable to read and write.
Read more here.