Big data is just possible, not everything.

The curtain of Brazil's World Cup has fallen, with millions of fans either happy or sad, or stunned or confused-the mood is complex but will be overcome, and life and work continue to move forward. Looking back at the overall game, perhaps some friends can remember, before the World Cup opening, a Yahoo American team had predicted that Brazil will eventually won the championship. Now that the dust settles, it is clear that the prophecy has not come true. However, this does not mean that the research methods of the relevant team are worthless.

The Yahoo research team drew on the football-related content from the 83.1 billion articles in a Tumblr 188.9 million blog accounts, and then focused on 27.3 million World Cup-related fan reviews from February to May this year, to "give each team a dominant value" approach, The Brazilian team is judged to have a larger winning side.

Unlike the great physicist Hawking of our time, who used the "World Cup formula" to measure the results, Yahoo's results were based on a discussion of light blogger fans, thus showing more of the inclination and anticipation of the majority of viewers-every fan has a champion team in mind, Samba football with strong strength and outstanding ornamental and popular, the title is very high. After Brazil's big defeat in Germany, the grief of the fans is a testament. But the results do not depend on the votes of millions of viewers, but on the performance of more than 10 of people on the pitch. As Prof Hawking says, football is much more complex than quantum mechanics.

In any case, Yahoo's forecasts are a very useful attempt. Similar research is helpful for companies to study markets and consumer demand in areas where fans tend to be strong enough to determine results.

Yahoo's analysis of the World Cup results could also spark more thinking: Can the big data that has recently been fired somewhat overheated predict the future? Some industry colleagues and analysts believe that data can reveal rules that help people and businesses predict results, while others argue that large data is limited and superstition is stupid.

As a researcher in this field, my point of view lies somewhere in between.

The big data reveals the correlation and the phenomenon, but not the law and the essence, the so-called "know it but do not know why" is the big data analysis of the results of the portrayal. A popular case in business schools, where data from stores shows the sale of beer and diapers. The study found that this was a tendency for young fathers to Sihuo beer when they were assigned to purchase baby supplies. If it is limited to discovering the association, the store may simply put the two items together, and the pattern behind the phenomenon will allow for more targeted promotions.

That is, big data can provide valuable clues, but they cannot substitute for artificial research-such as digging into the ground to uncover the hidden logic behind the consumer-behavior chain. Interested readers can see the book "Brand Brainwashing", the author is a veteran marketer, the book lists a number of vivid but perhaps surprising examples: modern marketing is so pervasive, for example, people have not been born marketing began, the women often go to the store background music to the baby has the effect of stop crying and so on.

These cases are often supported by research data, although not necessarily at the level of large data, but have been quite sophisticated, including the use of MRI scans of the brains of the testers.

People who are extremely optimistic or pessimistic about big data are in fact regarded large data as an extension of the traditional marketing model. Optimists are eager to find a "big kill" to achieve the perfect "lure" and control of consumers. But the pessimistic faction is more rational some-extremes meet, the excessive marketing will incur the consumer revolt, uses the big data analysis result to strengthen originally already like the mercury to flood the all-pervasive marketing, this is really good?

In my opinion, consumer behavior is difficult to predict and control---------------------------------------------- Publishers were surprised to find that the original war period of metal was ransacked, vendors lack of weight, with a variety of goods replaced, an unknown vendor found that the book weighs just a pound, so vendors generally buy to act as weight.

In this story, large data analysis cannot accurately predict this application situation. But we can do it, when the consumer produced in the book when weight such a fantastic idea, big data can immediately recommend to him "existence and nothingness." Since consumers are difficult to predict and control, on the one hand, the use of large data to observe and summarize the specific situation of the group behavior, on the one hand, more humble and serious communication with consumers, with cautious and persistent attitude and better products, services to sticky customers, rather than only in marketing efforts.

    In summary, large data can greatly improve the accuracy of predictions, it can only make products and services in the excellent enterprise become stronger, but not to save those who have fatal defects in the fire.