The difference between the large data age and the sampling analysis

In the big data age we want the whole data, no longer the sample data. The sample analysis has been developed for less than 100 years and has solved some specific problems at the time when the data and technology are limited. In the case of a census, the general statistics take 8-10 years, and at this time the data are completely inaccurate and ineffective. Therefore, a statistical method of sampling was proposed under the condition of the time.

Sampling analysis has its inherent characteristics: absolute randomness and neglect of detail observation. The purpose of the sampling is to obtain more information with minimal data. Absolute randomness: We need to be absolutely random when we sample, but it's difficult to do this, because people always have different opinions about the same thing. Deviate from randomness, the error rate of sampling result will increase greatly. Neglect of detail observation: the role of sampling in the macroscopic field has been lost in the microscopic field. As with the economic fringe theory, when the sample reaches a certain value, the information on the individual becomes less and fewer.

There is a very important point in statistical sampling analysis: The accuracy of sampling analysis increases greatly with the increase of sampling randomness, but it is not related to the increase of sample quantity.

When we can now collect http://www.aliyun.com/zixun/aggregation/13584.html "> Mass data, the sampling is meaningless to us." Collecting data in large numbers is not something that only a large company can do, and many companies can do it.

Large data refers to the method of using all data without random analysis. The advent of cloud computing, let us collect the massive data provides the infrastructure, through cloud computing to the Big data analysis, forecast, will make the decision more accurate, release more data hidden value, play a greater role.

This article is written by the sharing of the Internet (http://www.hed236.com), reprinted please be sure to leave a link, thank you!