Professor Hansheng of Peking University: Big Data is deified

Recently, large numbers have been mentioned and become a popular concept.

Companies are claiming their big data capabilities, but the "precision ads" that netizens are pushing are often useless spam information. What about the big data capabilities of Chinese companies? Big data facilitates life, and it brings privacy and security risks, where is the border?

June 12, on the hot issue of large data, Beijing News reporter and Hansheng, professor of Business statistics and econometrics at Guanghua School of Management, Peking University, opened a dialogue.

Hansheng

Professor, Ph. D., Department of Business Statistics and economic metrology, Guanghua School of Management, Peking University. Director of Business Intelligence Research Center, Peking University. Chief scientist, Boya Cubic Technology Co., Ltd. The founder of the "Bear" of micro-credit. Graduated from the Department of Probability and Statistics, University of Mathematics, 1998, University of Wisconsin-Madison in 2001. American Statistical Association 2014 (fellow). He mainly studies the high dimensional data analysis, the statistics in the electronic commerce domain application and so on, especially pays attention to the network data and the position track data statistical analysis.

Core point of view

Large data analysis is not a novel concept, the challenge is to improve the ability of large data with the new data types produced by technological progress, including Chinese text, network structure, location trajectory, etc., need to establish a new analytical model; In the domestic industry, the marginal profit-less electric dealers use large data to develop space, But traditional manufacturing has a good prospect of using big data, and the state needs to regulate the privacy risks that big data poses and balance the interests of industry and individuals.

Big data is more like a slogan.

Some companies are fashionable, emphasizing that they are big data, but in fact their data analysis ability is miserable

Beijing News: Recently, big Data concept is hot, many companies including listed companies are talking about big data. What do you think of the concept of large data?

Hansheng: This is not a rigorous academic definition. It is more like a slogan, a need for public propaganda. With the progress of technology, large data has a certain substantial change. For example: A new data type has been generated, reaching a certain level of magnitude. But there are also many deified places. For example, the original financial investment data is very large, but also in the practical application of the embodiment, but no one was concerned.

The Beijing News: Many companies now claim that their data have reached a new level of magnitude.

Hansheng: Some enterprises used to do logistics, some do 3C, now they are in the fashion, stressing that they are big data, but in fact, their data analysis ability is appalling. What it used to be, what it is now. Of course, it does not rule out that there is a good business, from beginning to start to focus on data generation value.

Large data development space for automobile industry

I am bullish on the traditional industries, such as furniture, cars and other good profit margins, and they have great space for future use of large data.

Beijing News: Now the domestic claim that the most powerful data is the power of the business sector. It is reported that the electricity dealers can now do the user login, they can judge what users need, so that the early delivery, the user wants to buy things to. Do you realize it in reality?

Hansheng: This is hard to achieve. For a handful of people with very regular buying behavior, their shopping needs are predictable. But in most cases, the consumer's buying behavior is highly unpredictable. Personalized recommendations exist for so many years, the transfer of goods into the customer's purchase behavior is generally a percentage of the conversion rate, if the 10% is already very high. After all, data analysis only describes the behavior of markets and consumers and does not help people make decisions.

Beijing News: Domestic electricity dealers now use large data mainly in the product page personalized push. How do you think you're doing?

Hansheng: The cost of the page push is very low, does not involve physical handling, its marginal cost is almost zero. In this respect the domestic do better and better, in individual cases conversion rate can achieve 10%. This process involves not only the accuracy of the algorithm, but also the overall quality of the site's services.

Beijing News: What is the space for future domestic electric dealers to further enhance their ability to use large data?

Hansheng: I am not optimistic about the electric business industry, because the marginal profit of the electric business industry is already very low. I am bullish on the traditional industries such as furniture, cars and other traditional businesses with good profit margins, as well as fund insurance. There's a lot of room for them to use big data in the future. Another big big data is the direction of the use of marketing-related, for small and medium-sized enterprises to solve the difficulties of advertising services.

Beijing News: How to use large data to help SMEs solve advertising problems?

Hansheng: Small and medium-sized enterprises do not have the advantage of online marketing, a only for a few kilometers inside the café customers do not need to go to the portal site or television to advertise, small enterprises can not afford such ads. They need to pinpoint ads. There is a lot of room for directional marketing with large data generated by lbs (location-based services) tools. But now the problem is, based on the user location of the marketing platform more and more, small and medium-sized enterprises screening costs very high.

"Convenience" and "privacy" need to balance

Privacy protection is too loose, Internet users ' privacy is not protected; too tightly controlled

Beijing News: Despite the huge potential of large data, but also to the personal and corporate information security risks.

Hansheng: Now our country, even in the global scope, the legal definition of privacy protection is not clear enough, lack of unified understanding. For example, Internet users in the electronic business to browse the record of goods, is the netizen, or the electric business, or netizens and the electricity quotient altogether have? There is no conclusion. European regulation of personal privacy is very strict, but it also limits the development of Internet companies in Europe. The United States has a relatively loose regulatory role in this area, and China is still learning to explore the stage. But privacy protection this knife where there is no conclusion, pipe too loose, internet users privacy is not protected, tube too tight, enterprise Innovation Limited, industry development is limited. Therefore, we enjoy the convenience of the Internet, but also need to let some privacy. However, the specific needs of the transfer of the number of countries, enterprises, individuals to understand the gradual understanding of communication.

The Beijing News: How accurate can the data analysis be to the person's recognition now?

Hansheng: I'm not sure about China's ability to analyze this. According to the published literature, 87% of the people in the United States can be identified independently of the postal code, sex and date of birth. At present, enterprises can through a person's purchase behavior, identify a unique virtual person, enterprises can know this virtual person's many preferences, but this person's name, do what, in general, the enterprise or not know. Ordinary enterprises, there is no incentive to know. But if you have good people, you can identify the specific person by docking the data that the dealer obtains with other data. So privacy protection is still very important.

Beijing News: Now smart phone installation software, many require access to a large number of permissions, and some even require the right to listen for calls and text messages, these software for the large range of personal data collection is necessary?

Hansheng: I don't understand the motives behind these actions. According to my understanding of the industry, most enterprises take back, nothing can do.