Four bits of credit card metadata is capable of re-identifying nearly 90% of the people, warns a study carried out by Massachusetts Institute of Technology (MIT) researchers.
For the purpose of the study, the researchers observed three months of credit card
transactions and records of 1.1 million people who shopped at 10,000 locations. Lead researcher Yves-Alexandre deMontjoye and his team studied the time, data, location and price of each transaction and not the item which was purchased. He did not look at the identity of the individual in question either.
The names of the credit card holder or any information that would tie back to the individual were erased. The researchers also replaced the card account numbers with ID numbers that were assigned randomly.
The credit card data was taken from a single bank in a country, which has not been named. The period studied was from Jan. 1, 2014 to March 31, 2014.
The purchases were selected randomly for each ID number and the researchers then analyzed the purchase history of other customers to determine who had the same data points.
“We were really trying to quantify how many pieces of information were needed,” revealed deMontjoye.
The researchers also diversified the data points given to each customer from two to five. They discovered that even sans price information, two data points were enough to clue in on the identities of over 40 percent of the people in that data set. Unsurprisingly, five points and price information was adequate to earmark pretty much every subject.
The results show that the “four spatiotemporal points” or metadata i.e. time, date, place and price of transaction were adequate to exclusively recognize the shoppers with 90 percent accuracy.
“We show that knowing the price of a transaction increases the risk of reidentification by 22%, on average,” reveal the researchers. “Finally, we show that even data sets that provide coarse information at any or all of the dimensions provide little anonymity and that women are more reidentifiable than men in credit card metadata.”
The study has been published in the journal Science.