Big data: A shift in computing and thinking

In the past two years, the big data has been widely discussed by the public, and even become the selling point of many business marketing. Undoubtedly, the development and popularization of intelligent devices make the mass data collection possible. But the big data is not pure "data big", it contains a kind of calculation and thinking way change, want to exert big data insight, still face the challenge of collecting, managing, analyzing data. How can these obstacles be broken? How can large data be used in the future to create greater value? These questions are worth our cool judgment in the heat of the big data.

April 26, Tsinghua University set up "Tsinghua-Qingdao Data Science Research Institute", while holding a large data era high-end forum. In the previous two days, Baidu in the fourth session of the technical opening day, officially announced the opening up of large data engine, to provide large data storage, analysis and mining technology capabilities. Does it mean that large data applications are entering a new phase, with large data being included in research and open engines for businesses?

Large data

Traditional statistical methods pursue precision, large data only predict macroscopic trend

This is a large number of technical concepts, now more and more like a marketing tool. From automobiles, cosmetics to sports, in the mouth of marketers, it seems that all industries can use large data, accurate positioning, find consumers, forecast trends, to win the future.

Guoming, professor of journalism at Renmin University of China, said that at present from the domestic situation, the real use of large data analysis of the success of the case is not much, many companies are large data as a marketing gimmick, the analysis is mainly based on traditional data analysis methods.

In fact, the industry does not have a unified understanding of how much data can be called "Big Data", and it is generally assumed that 100TB (too byte) is the threshold for large data. In short, the data that the traditional method cannot handle is large data.

Large data is generated by the development of mobile internet and smart phones, all kinds of smart wearable products, people's behavior, location, and even physical characteristics of the body can be easily recorded data, which makes the massive data acquisition possible. In fact, the current data acquisition volume is showing a rapid growth trend. A new forecast by an international data-statistics agency suggests that in 2020, the amount of data generated worldwide is expected to reach 40ZB (Chak Byte, 1-kilobyte equals 1 billion-byte).

But large data cannot be simply understood as large data. Huaijin Peng, a large data research expert and president of the Beijing Aerospace University, said that the large data has "large scale, fast change, kind of miscellaneous, low value density" Four characteristics, is a challenge to traditional computing and thinking mode.

First, because almost every data point can be collected, comprehensive data instead of sampling, one-sided, partial data. "For example, traditional sampling, we need to ' taste ' and ' taste ' the sample data in the beginning and the middle, but in the big data age, the random sampling method may be ineffective. "The bosom enters Peng to say.

Bosom enters Peng thinks, because the data measurement ability is limited in the sampling analysis, the statistic pursues is accurate, want to obtain the most information with the least data. But the big data is messy, the complete precision does not exist, also is no longer pursues the absolute goal, the big data only needs to make the rapid forecast to the macroscopic trend.

Another change is to relate from a focus on causal shifts to data. In the big Data age, "the reason behind the data is no longer important, people just need to know that there is a statistical correlation between the data." You need to know the reason why. "The bosom enters Peng to say.

In the view of proponents of large data, data has been able to speak for itself, the traditional scientific statistical model is outdated and the theory may be terminated.

Big Data marketing is mostly gimmick, some organizations can't even collect massive data

As the author of the Big Data age, which is known as a major data system study, the authors point out that big data is a new ability of society: in an unprecedented way, through the analysis of massive data, to obtain great value products and services, or insights.

Large data contains the discovery of facts, mining value, predicting the future insight, but also a wide variety of large data marketing theory starting point. In fact, big data insights do come into play in sectors such as public health and transportation.

Fu, associate director of China Center for Disease Prevention and Control and academician of Chinese Academy of Sciences, also shares the role of large data in public health prevention and control. He says large data can provide some explanatory information before the flu arrives, providing a cushion against flu prevention.

Similarly, in the era of intelligent transportation, mass vehicle information can not be analyzed by the traditional way, but with the help of large data, it is possible to predict the future flow of traffic, road, etc.

"The big data that you can talk about" is really as good as the marketing staff?

Analysts point out that while data storage and handling are becoming more convenient, large data applications now face the challenge of data collection, management, analysis of massive amounts of data and creating value.

"If the data compared to books, the number of books, the first to find the storage of large data of the ' Big library ', the next step to solve the problem of data query, there is no good query engine, the book can not find, the data is difficult to use." Baidu Big Data director Li Gangjiang said. The reality is that most organizations and businesses do not have the capacity to collect and analyze data.

Industry insiders pointed out that large data in some areas of marketing is only gimmick, regardless of the large data analysis results are effective, some industries even the basic large data acquisition and management conditions are not available, let alone accurate positioning and forecasting.

Baidu Senior Vice President Wang Jin also said that the traditional database does not manage the capacity of large data, traditional industries how to enter the era of large data, the use of large data value, is placed in many industries in front of a new topic.

Increased computing power and reduced cloud storage costs will benefit large data technology changes

Baidu chief executive Robin Li believes that with the rise in computing power and the cost of technology products such as cloud storage, big data has come to the tipping point of technological change. Not long ago, Baidu launched a "Big data engine Baidu", Baidu hopes to use the tool to collect, store, compute, excavate and manage large data, and through deep learning technology and data modeling technology, so that the data has "intelligent" technical capabilities, service of traditional industries.

According to understand, Baidu Big data engine includes open cloud, Data Factory, Baidu brain three big components. Among them, the open cloud solves the problem of data storage and computation, the Data Factory standardizes the industry data, provides data management and analysis, and the "Baidu Brain" lets the machine think like the human brain and analyze and process the data.

However, analysts point out that while many tools have been developed for large data mining, the mature application of large data is still a long time. First of all, data clutter, low value density, how to effectively collect data information is still not mature scheme. At the same time, the size of the data does not determine everything, regardless of the type of data analysis, there may be statistical shortcomings, can not say that the data is larger, updated, faster without problems.

Wugansha, chief engineer of the Intel China Research Institute, said that large data, as a new form and practice of data, would enrich the methods of data application, but could not replace the traditional statistical analysis methods, and not to hallow large data.

Many hand rings are said to collect personal health receipts.

Wear a watch, hand ring daily Body Index

What is the purpose of buying wearable equipment? New fun, sports socializing, or managing personal fitness habits? In fact, these do not play the real value of wearable equipment. It is understood that the market today, many wearable products are claimed to be able to integrate wireless networks, mobile computing and automatic identification, including blood sugar, heart rate, respiration frequency, weight, hydration and body movements, and other physical indications, can be understood in real time, this is called "Large data medical", so that many more and more health-conscious white-collar workers for their own, Families to acquire these products.

Although IDC forecasts that China's big data market will grow 5 times times in the ~2016 years of 2012, the government, banks, health care, telecommunications and other industries will occupy the largest share, but "big data medical" in the domestic real landing still a long way to go.

Market

Many manufacturers at home and abroad devote themselves

According to foreign media reports, Apple is rapidly expanding the medical team, recruiting fitness experts, medical equipment industry experts and other medical sensing background engineers, and is likely to release Healthbook this fall. It is reported that Healthbook can track from sleep to nutrition, A variety of indicators from movement to vital signs, including blood sugar, heart rate, respiration frequency, weight, hydration and physical activity, have become the key selling points for Apple's next-generation mobile operating system, iOS 8 and its rumored iwatch smart watch, and have become the tipping point for the mobile healthcare industry.

In fact, health and sports applications have become increasingly popular in recent years. The Azumio Company's health monitoring and fitness applications for Apple's iphone have reached 40. In the opinion of it bosses, not only is the medical data mobile collection, the formation of "big data" has more imagination space. Currently, the Windows 8 system integrates Bing Healthcare (Bing Tiyatien &fitness) to help users record exercise, medication, and diet, and Microsoft's Healthcare data platform HealthVault allows users to collect and manage health and physical condition information about themselves and their families. Combined with wearable products Fitbit or Nike Fuel Band data collection provides more convenience for medical care. In addition, the United States fourth largest electronic medical record service provider Practice Fusion recently and for smartphones do heart rate monitoring accessories AliveCor to work together, the equipment in the data will be integrated into the medical records, and stored in the cloud for immediate access.

"Once enough data and samples have been accumulated and put in the hands of professional medical personnel, it will be revolutionary to push the health care industry forward," he said. "The Intel Institute is also working on the interconnection of medical data," according to Wang Yichun, director of partner relations at Intel's Software and services division, "to allow small to simple pedometer, large to complex CT scanners to connect to each other, and to communicate and share data with the cloud." "In the large data medical background, the human body signs can be monitored continuously, the doctor is no longer only after the illness of doctors" Hango.

Problem

The data collected by the manufacturer lacks certification

At present, the domestic "Big data medical treatment" can only be called "tele-Mobile medical treatment", some even just facilitate the wireless data transmission inside the hospital. For example, Medex's "Synchronous handheld ECG machine", although it can be used for portable tablet computers, but patients at home by themselves not convenient. Other domestic manufacturers are also in mobile care, community medical services, surgery anesthesia, ECG monitoring, clinical services and other fields to provide a variety of solutions, but the quality varies, technical is also separate, making wearable equipment data is extremely dispersed, equipment to generate a key data, can not be used by other agencies, but in fact, Health-related data needs to be widely used to be useful.

In addition, if health-related data are to be used for medical purposes, data must be certified and licensed by the Government's drug monitoring agency. The majority of vendors are not certified, and the data itself is due to the quality of the sensor has an unknown error.

IBM Medical Business Development Manager Liu Jingwei that the medical industry's information characteristics and many other industries, there are many semi-structured and structured data, and distributed in different medical institutions, so how to effectively integrate it is a challenge. "Every patient rarely goes to a hospital. One of the goals of large data therapy is to effectively integrate data related to each patient's health, to use evidence-based medicine and digital-driven analysis to see risk-related factors, and then to plan and implement them effectively. But such integration is very difficult. ”

Experts

Need to unify the standards

Wang Cai, deputy director of the National Health and Family Planning Commission's Statistical Information Center, said: "The ' Big Data medical ' has brought immeasurable value to our medical process reengineering and medical efficiency improvement, but it will take time to move to a truly mature application." ”

At present, the information function of medical institutions is strong, but the overall synergy effect between medical institutions is poor; the vertical health service system is strong in function, but the standardization construction is weak, information sharing and business cooperation mechanism are scarce among the systems, and there is no interoperability between the system. "Experts believe that with the development of mobile medicine, different medical institutions in accordance with their needs to deploy customized mobile solutions, medical industry has become the first to start large data applications, one of the pioneer industry, large data, virtualization and other technologies to support the application of mobile medical end.

"Various types of hospitals, community health service centers, rural medical workstations, disease surveillance centers, emergency centers and other health care institutions are dispersed." In the development of medical informatization in more than 10 years, the IT system under deposition has many technical categories, which brings great challenge to data collection, quality, data standard and subsequent maintenance. If all uses the custom development, the standard interface or the manual entry way, inevitably must invest the massive manpower and material resources, and in the data accuracy, the real-time sex cannot be guaranteed. ”

(Responsible editor: Lu Guang)