What is the difference between big data and BI?

We have always kept a principle in mind in the implementation of the project: there are many opposites between big data and BI: for example:
　　 1. Full thinking. Without sampling modeling, go back to DW implementation. Big data uses the full amount of data to run directly. First, find the fields in the large wide table that were previously selected by business operations, that is, rely on machine learning to build rules, and then implement them in the global data. At this level, big data is first and foremost infrastructure, capable of accomplishing tasks that could not be accomplished before. For example, in an insurance company, SAS can only run a classification model of hundreds of thousands of users. Now it can run the full amount of 36 million users of individual insurance, and get thousands of classifications. It is found that business operations cannot give facts. This is the IT attribute of big data, the subversive innovation brought by Hadoop distributed computing.
　　 2. Personalize. BI is decision-oriented and human-oriented intervention. The output is more in the form of dashboard.report. Therefore, the description of facts is more based on group commonality rather than individual characterization. For example, going back to the example of an insurance company, when we use big data to calculate the probability of churn risk for each individual customer and make a personalized customer view. The BI system needs to aggregate into macro statistics. The former helps us deeply understand each user, and is suitable for precise recommendation marketing questions and answering vague questions about how much how strong; the latter helps decision makers grasp macro statistical trends, and is suitable for business and operation indicators to support questions and answer accurately Statistically significant probability question for yes or no.
　　Big data's description of individuals or group descriptions of BI leads to the third difference below. Recommended understanding: Yonghong agile bi
　　 3. Insight or Automation group common description of the obtained Yes or No question . It is becoming more and more difficult to answer complex economic phenomena, and there are more and more dimension indicators. For example, in the above example of customer churn risk, the BI system is used as a statistical indicator and then reported to the senior management for decision-making. Executive strategies such as the formation of customer retention plans by senior management are often very risky, including questioning the accuracy of data, execution efficiency feedback cycle, etc. The result is that in the face of such macro-level inaction and inaction. Therefore, in a sense, the Insight formed by BI did not play a role because there was no closed-loop intervention.
　　Big data emphasizes that Automation Taobao also emphasizes the production system that automatically recommends each online purchase. In the above example of the loss of insurance customers, big data depicts customers and provides them to front-line sales, data services sink instead of summary reporting, each insurer makes micro-decisions and micro-actions, and gives timely feedback on small risks. Big data puts more emphasis on providing automated tools rather than statistical reports.
　　 4. Feedback and experimental methods. Because of the introduction of Automation and the recording of behavior data, the effect is more direct and timely. For example, AB testing. Hulu, which is the most used on the Internet, has more than 200 tests online every day, which shows the importance of testing methods in data analysis.
　　I have the opportunity to talk about BI and big data. In this era of big data hype where concepts are flooded, various companies transform big data overnight. How to distinguish between small data and big data is not only the technical foundation, but more importantly, the way of thinking.