1) data quality. Divided into data standards and data accuracy. Noise in data should be eliminated as much as possible. For the quality of data, a lot of manual work is indispensable.

2) Business scenario of data. We can't do all the scenarios, so business scenarios and product forms are very important. Personally, the narrower the business scenario, the better.

3) The results of data analysis should be understandable and know what to do next, rather than data for data's sake.

There are many people engaged in data mining, but there are not many successful cases (compared with a large number of attempts). At present, I seem to think that the current data mining technology is a transitional technology and still in the exploration stage. In addition, many data mining teams are unprofessional and unskilled, and feel sorry for the technicians. ...