Big data forecasts from Brazil fiasco

This 4 o'clock in the morning, Germany's World Cup finals for Brazil gave everyone a big bang. Germany's 7:1 victory over Brazil is breathtaking, and Brazil's fiasco is hard to get. Perhaps even Google's big data forecasts do not predict a 7:1 disparity.

In this case, it may be a bit far-fetched to discuss using large data to predict the world accurately, but after all, big data predictions are trends.

Big Data today, although there is still a distance from the perfect prediction, however, it is undeniable that the belief in data is more reliable than the belief in intuition. Aside from this "Big score" game this morning, Google, Baidu, Microsoft and so on, through the analysis of large data on the World Cup's early predictions are equally surprising.

"Success" predicts World Cup top 16?

Google's cloud computing platform has successfully predicted the winners of each of the top 16 World Cup games. It is understood that Google's use of real-time sports data company Opta QSL data, as well as the BigQuery engineers Jordan Tigani developed the strength of the ranking system, but also take into account the degree of enthusiasm of the audience data to calculate the home advantage, so as to predict the results. In addition, Google used the system to predict the World Cup 8, the result is surprisingly accurate: Brazil to Colombia, Brazil wins the probability of 71%, France to Germany, France wins the probability of 69%, the Netherlands to Costa Rica, Holland wins the probability of 68%, Argentina to Belgium, Argentina wins the probability of 81%.

In fact, Google is not the only company to make perfect predictions for predictions, and Baidu and Microsoft Bing have also made predictions, and what we all have in common is to make predictions based on a comprehensive analysis of cloud data systems.

With the development of large data industry, Google, Amazon, Ali, Baidu, Tencent, because of a large number of user registration and operation of information, naturally become large data companies. Records of various data may seem random, but when analyzed by the speed of light computer, they reveal images, patterns, connections, and trends, not only to improve business performance, but to change life.

Search engines, such as Google and Baidu, not only store the network connections that appear in search results, but also store users ' search-keyword behavior, which accurately records the time, content, and manner in which people conduct their search, predicting your intentions before you realize what you're looking for.

Guess the examination questions, the epidemic prevention situation big data is omnipotent?

During the Spring Festival last year, Baidu has begun to forecast the trend of population movement in the Spring Festival; this year's Qingming and 51, Baidu to the country's major scenic spots, the city's flow of people to do the forecast, the 2014 college entrance exam composition Proposition direction of the prediction is "hit" the national 18 sets of questions It is understood that Baidu's "college Entrance Examination prediction" can also use historical search data, years of admission scores, various batches of provincial control line to predict the application of various universities nationwide, difficulty, the trend of various professional candidates and the province of the candidates are interested in what professional, school and so on. Baidu CEO Robin Li said, "The data mining is just the primary stage of large data technology." In addition to analyzing laws and trends through large data, machines must be able to think for themselves. ”

In addition to the IT business plan for disease prediction, real estate forecasts, employment forecasts, financial forecasts, China's CDC also plans to use large data, the early identification of a certain scale of unknown diseases, for the control of the epidemic time.

However, from the current point of view, large data analysis and prediction ability is far from perfect. A few weeks before the swine flu outbreak in 2009, the "Google Flu trend" predicted the spread of influenza in the United States, and its findings were even specific to specific regions and states, and were so timely that public health officials were shocked. In 2013, however, Google's predictions of flu were almost as much exaggerated as the results of the US Centers for Disease Control.

Industry believes that the future of "large data accurate analysis not only depends on the expansion of data resources, but also based on the development of large data engines." "It is understood that IBM has launched a large data industry solution, Intel shares the large data start-ups Cloudera, but also launched a hidoop based" large data engine.

Experts：

Data synergy and privacy issues to be solved

Google, IBM, Oracle, SAP and other enterprises in the field of large data technology innovation, more and more foreign enterprises by virtue of technical advantages and advance experience to enter the large data market. However, the development of China's large data industry is still in its infancy. "Every click, Touch, SMS, micro-mail, micro-blogging, driving, flight, call, photo, purchase, etc. all produce data ... Although a lot of data is produced every day, it does not show enough power. "The analyst at Sadie said," The Transportation department has the big data of the vehicle networking, the thing networking, the network monitoring, the ship networking, the Wharf station monitoring and so on, the health department has the influenza legal report data, the national influenza-like case sentinel surveillance and the pathogen monitoring data, the Public Security department has the massive video surveillance data, But government departments have almost no large data processing and mining technology. ”

In addition to Internet companies, Wal-Mart, China Mobile and other traditional enterprises also have a large number of user data, platform enterprises to use each other independently of the data gold, to get everything, but the private possession of data seriously restricts the wide application of large data and the integration of development. "Large data coordination can achieve intelligent path planning, capacity management, influenza prediction, vaccination guidance, security and so on." ”

The big Data age says, "the big data itself is a trend, not a precision, and to be infinitely close to the statistical results, it is necessary to complement the big data with the sophisticated traditional statistical methods, rather than the substitution of the two." ”

In addition, the privacy of the data remains to be solved. Only 4% of cancer patients involved in the Cancer Prediction program, which Google invested heavily in, have taken part in the clinical Trial database program, which means that up to 96% of the patients ' medical and physical signs are difficult to learn from other medical institutions or physicians.

Part of the content is excerpted from Guangzhou Daily