Big data is probably the most fashionable word in almost a year.

Big data is probably the most fashionable word in nearly a year, and the real essence of big data is not "big", but a whole new set of thinking that is connected to the Internet. What is the biggest difference between big and traditional data?

1. Online. First, the big data must be always online, and the online is also hot backup, not cold backup, not on tape, is ready to call. The data that is not online is not big data, because you have no time to export it to use. Only online data can be computed and used immediately.

2. Real time. Large data must be reacted in real time. We enter a product Taobao, backstage must be in 1 billion items, instantaneous rendering. If I have to wait one hours to present the speech, I believe no one will be on Taobao. 1 billion of goods, millions of sellers, 100 million of consumers, the instant completion of matching rendering, this is called large data.

3. The whole picture. Large data also has one of the biggest characteristics, it is no longer a sample thinking, it is a whole thinking. When it comes to data, the first reaction is to sample, sample, but the big data is no longer sampled, no more calls, we want all the possible data, it's a whole picture. Actually, the whole data is more accurate than the big data.

This is the three essence of large data, online, real-time, full picture.

3 Typical characteristics of large data

In order to have more understanding of the big data, I will discuss it with you. We do business, the most easily thought of two data applications, one is market research, send a marketing company or marketing department to do a survey, to see what the company feedback. The second is business intelligence bi, data mining, viewing data management reports. This is the most traditional two data applications. There are several typical features of such data applications:

1. Be aware of the objectives to be achieved and take the initiative in collecting these data. Because the computing power of each enterprise is different from the cost, how much time and what data can be used by the data will vary. And the large data is real-time record data. In principle, any person on any website, do anything, everything will be recorded, no personnel to do first distinction. So everyone stops asking, the data is recorded, so this is the first difference.

2. The person involved is no longer conscious participation, but unconscious participation, and you are doing things for your own benefit. You use a search, you are involved in Google's big data collection, because each click is a data source. If you take part in a market research, 80% of the cases you will refuse, 15% of the circumstances you may ask for a sense of compensation. Few people are willing to take the initiative in market research because it is a burden to you. But the big data on the line is an unconscious, self profit behavior for most people. I taobao is to buy things, I on Weibo is to see the news, I on Baidu is to search, you are for their own interests triggered by an unconscious behavior, but this unconscious behavior, have contributed to large data.

3. One is one-way, one is two-way. The data analysis we used to do was to assume a purpose, and then get the ready-made data, analyze the behavior, to test my guess. These are a one-way led. Large data in nature must be two-way, like search, when you click on the search engine click, you are to give it input data, it gives you the result is that it interacts with you, is that it brings you the value of the data. The big data itself is at any time creating value for you, so it becomes a two-way interactive positive loop in which both parties contribute data value. Any large data application, if not in the design of this two-way, mutually beneficial positive cycle, is not run, in essence, is not large data.

Large data applications, the reaction rate is the key

Finally, I would like to emphasize that the rate of reaction-the greater the data value of large data, the higher the rate of reaction will be. For example, Google's search, you enter a keyword to see the results, and one hours later enter the same keyword results, it is likely to be different. Because it has recalculated all the clicks in the world within one hours, then the information is optimized and then fed back to you.

So, you can think about it, the faster the feedback, the greater the value it creates, the greater the incentive for consumers to participate. The bigger the data, the faster the reaction, the better the result, the more users will become a black hole effect. This is the core concept of the Big data I want to talk about.