Big Data predicts no World Cup

God laughs when man thinks, and the World Cup comes, and God smiles and his blood pressure rises. It's not how much fun football is, it's the World Cup predictions.

I have summed up a little bit, excluding Bailey and Colonel's famous contrarian index, the World Cup forecast can be divided into five categories: first, the power of determinism, Goldman Sachs and the bank and other investment banks as representatives, will http://www.aliyun.com/zixun/aggregation/ 19351.html ">fifa rankings, past record, recent performance, player status, betting odds, national economy, etc. a series of data reflected the strength of the team as a basis for forecasting, using the econometric model to calculate the probability of looking extremely accurate, but the results of the calculation is very different, for example, Goldman Sachs predicts Brazil's title, while Deutsche Bank is more bullish on England.

Second, conditional determinism, represented by Professor Hawking, the effects of various objective conditions on a team's performance are leelawadee into a winning formula that includes course selection, referee nationality, shirt selection, formation selection, and vehicle choice, but the formula has a short forecast and is so far away from the final. The formula also predicts no final title.

Third, gossip determinism, which is represented by Yahoo and Microsoft Bing, converts the network noise in billions to a clear result signal. Although this kind of Internet magic appears to be both scientific and mysterious, and can even be explained from the theory of "expected self-realization" of economics, it is always reminiscent of the emperors ' special attention to nursery rhymes and bards ' works in ancient times, both in China and in the west.

Four is the fatalistic attribution theory, with a about 1966 and 3721.html ">2014 years of space and time overlap Hot post for the representative, 1966, the Champions League champions are Real Madrid, Spanish champions are Atletico, West second champion is the Betis, Serie A champion is Juventus, the downgrade of West B, This is true for 2014 years without exception. "Yesterday once more" left the World Cup, 1966 is England's title, according to Fate, 2014 England will also be the winner.

Five is the animal prophet, Paul as the representative of the Octopus in 2008, the European Cup 6 5 wins, 2010 Paul forecast South Africa World Cup 8 victory, but, Paul did not live to the 2014 Brazil World Cup unveiled, so the tortoise, elephants, pandas and other animal prophets on the projection stage, but, Paul's brilliance is harder to replicate.

Of course, all of these World Cup predictions are more or less part of a spoof, not true. But a little serious, detailed analysis of these five methodologies, in addition to the last one, the remaining four kinds of use of large data, this is the current large data heat of a performance. From the World Cup forecast to see the big data heat, three misunderstandings deserve to be alarmed.

First, big data doesn't solve all the problems. Large data is very useful, almost every link, including product design, market positioning, target selection, sales strategy and model correction, can rely on large data to solve the problem, even a variety of business-related issues, can also be from large data here to get help. But the big data is not omnipotent, the big data can solve is objective problem, not subjective problem. For the World Cup forecast, who can win the championship depends on a variety of factors, the various forecasting methodologies are essentially the same, except that they give different determinants of the weight is not the same. That is to say, decide what is the key element of the championship, this is a subjective problem, the large data can only explain the objective state, can not be mixed with important subjective judgments directly to answer the question.

Second, the data is not as big as possible. Data analysis has long existed, big data is hot but the recent thing, the difference is in a "big" word. In principle, the more data species, the longer the sequence, the richer the information contained. But it is worth emphasizing that the richer the information is not necessarily the better, now is an information explosion era, the more information sometimes the more adverse to learn. And more importantly, like the World Cup forecast, is obviously a core information is not prominent things, so information is easy to form a logical conflict, this is the prediction methodology is numerous, who are not reliable reasons. In addition, the larger the data, the more choices in the data processing process, and the choice is always difficult to avoid moral hazard issues, each large data analyst, have to face the temptation to choose for a specific goal, "selective choice" is often large data but blurred the underlying reasons for the truth.

Finally, the large data are uncertain Schrodinger's cat. Before the box opened, Schrodinger's cat was both alive and dead, and only the act of opening the box would lead to a collapse of state superposition and uncertainty would disappear. In other words, before the World Cup really started, 32 teams were champions, not champions, only to kick up to see the real chapter. Large data analysis is based on fait accompli, and it is better at telling us what the world is, not how it should be. In the face of uncertainty, large data is weak, so the fundamental, large data is not used to predict. What big data can bring to us is knowledge, not judgment, not arbitrariness, but respect for uncertainty is part of the proper use of knowledge.