Ten Commandments of Big Data Business Intelligence
"This is my last version," this is a visual map, copy it, break it, this is what I get these insights, see what can be used in. This will help others learn something useful from your perspective.
The fifth commandment: to the natural form
Big data is unstructured, and we have heard too much about it. It's not true. Hong Kong server Finance and sensors generate a large number of key pairs. (It may be the most popular data format at present). It can be semi-structured, multi structured, etc., and the database focuses on this data format. It has the advantages of good processing and scaling, but if it is converted to a form, the expression will be lost. A large amount of big data is still made into tables, usually with thousands of columns. You must find relationships for all values: "In this case,"... choose "" from here. Flattening destroys the key relationships expressed in the original structure. Far away from the solution, tell you, "Please convert the data into tables, because we have been doing all this time.".
Sixth commandment: Don't wait for the result indefinitely
, we expect data processing speed to become faster. A typical method is multidimensional data set, which essentially transfers data to a pre calculated cache to speed up processing. The problem is that you have to extract and transfer data (see the first commandment) before building a data cube, you can speed it up. Now, this method works well on a certain data scale, but if your temporary form is too large, your laptop will crash when trying to localize. When you pick up new data and rebuild the cache, stop analyzing the new data. In addition, take care of the sample problem. You may get a good looking, good looking point of view, but eventually you will find that the whole thing is wrong. The problem is the lack of a holistic perspective. You can easily adjust the data by selecting a tool.
Seventh commandment: do not report, but for a long time
Create Application
