Ů��С��Ƶ

Baidu, China’s internet search giant, has shown just what you can learn when you have access to enough location data.

The firm’s Big Data Lab in Beijing has announced that it has used billions of location records from its 600 million users as a lens on the Chinese economy, tracking the flux of people around offices and shops as a proxy measurement for employment and consumption activity. The lab even used the data to predict Apple’s second quarter revenue in China.

We already know that location data is useful, tracking population movements and the spread of disease, for example, but this is the first time that a company on the scale of Google, Facebook or Baidu has shown its hand. The data generated by their huge user bases gives these companies enormous power and insight that they don’t typically talk about. Academic researchers have great difficulty accessing databases like this. But Baidu can just peer into its own servers. The search giant is saying exactly what it can do with the data, and how much data it has.

First, the researchers hand-labelled thousands of areas of interest – offices, shopping centres and industrial zones – across the country. Then they studied the location data – which runs from the end of 2014 to the middle of 2016 – to see how many people were at those places at each time, and how that changed through the year.

Bird’s eye view

The data presents a strange bird’s eye view of Chinese society. It captures a large shoe factory in south-east China closing down in early 2016 – the area of the factory, once buzzing with location traces, suddenly empty like an abandoned beehive.  A few months earlier, a busy mobile phone factory in Jiangsu province, on the east coast, is gutted.

The data also shows success and growth. The number of people going to a software park in Beijing doubles from 2014 to now, while attendance at a start-up’s offices skyrockets after it receives investment in mid-2015.

Baidu has collated all the data to build an employment index for China, a number that reflects the overall state of the labour market by tracking how many people are visiting industrial, manufacturing and technology zones in the country. The index shows that employment in manufacturing has dipped by roughly 10 per cent in China since 2014, while high tech employment has grown slightly.

Apple revenue

Baidu has built a similar index for monitoring consumption activity, measuring how shopper footfall changes over time. It also takes a crack at predicting Apple’s Chinese revenue for the second quarter of 2016. They tracked how many people visited Apple Stores in China over the first quarter, when Apple’s Chinese revenues are known. They then compared foot traffic for the current quarter to predict that Apple’s Chinese revenue has fallen by 20 per cent.

“To the best of our knowledge, we are the first to measure the second largest economy by mining such unprecedentedly large scale and fine granular spatial-temporal data,” the Baidu researchers write.

Yves-Alexandre de Montjoye, at Imperial College London’s Data Science Institute, says it’s good that Baidu is being open about its impressive capabilities.

“It’s definitely an encouraging trend to see them being open about the research they’re doing,” he says. “After all the controversy about the Facebook emotion contagion study, one bad result would have been to stop seeing this research being published.”

De Montjoye says one issue still remains – whether we really want just a few large web companies to control the power they wield through this data. “It really asks the question of how can we make this data more available, both in the NHS and Google’s case and here,” he says.

Read more: Internet 3.0: How we take back control from the giants

Journal reference: