Who can talk about the differences among data warehouse, data mining and BI and relevant suggestions?
The input end of data warehouse is different data sources, such as: some data are stored in mysql, some data are stored in mongdb, and some third-party data sources. The final output is used for enterprise data analysis, data mining, data reporting and other directions.
Multiple data sources are extracted by ETL(Extract- data; Conversion-data conversion; Load-data loading) integration.
There is also the relationship between data warehouse and database. Here, I found a picture on the Internet, which is still relatively clear.
Business-oriented database is usually called OLTP, and analysis-oriented data warehouse is also called OLAP.
Data mining: Data mining can see through your needs. Broadly speaking, any process of mining information from a database is called data mining. From this perspective, data mining is BI. But in technical terms, data mining refers to cleaning the source data and transforming it into a data set suitable for mining. Data mining completes knowledge extraction on this fixed data set, and finally makes further analysis and decision using appropriate knowledge patterns. In this narrow sense, we can define that data mining is a process of extracting knowledge from a specific form of data set. Data mining often chooses one or more mining algorithms for specific data and specific problems, and discovers hidden rules under the data, which are often used to predict and support decision-making.
Related sales cases:
Supermarkets in the United States have such a system: when you buy a car full of goods, after the salesgirl scans your goods, some information will be displayed on the computer, and then the salesgirl will kindly ask you: We have disposable paper cups for sale on the F6 shelf. Do you want to buy it?
This sentence is by no means a general promotion. Because the computer system is ready, if you have napkins, large bottles of coke and salad in your shopping cart, there is an 86% chance that you will buy disposable paper cups. As a result, you said, ah, thank you, I haven't found the paper cup just now. This is not a magical scientific fortune-telling, but a system realized by using the association rule algorithm in data mining.
Every day, new sales data will enter the mining model, and together with the historical data of the past n days, it will be processed by the mining model to get the most valuable association rules at present. Using the same algorithm to analyze the sales performance of online bookstores, computers can find the correlation between products and the strength of correlation.
After several years of accumulation, most large and medium-sized enterprises and institutions have established relatively perfect basic information systems such as CRM, ERP and OA. The unified feature of these systems is that the database is finally added, modified and deleted through the operation of business personnel or users. The above system can be called OLTP (On-line Transaction Processing), which means that after the system runs for a period of time, it will certainly help enterprises and institutions to collect a large number of historical data. And a large number of scattered and independent data in the database are just incomprehensible gobbledygook for business personnel. What business people need is information, which is abstract information they can understand, understand and benefit from. At this time, how to transform data into information, so that business personnel (including managers) can fully grasp and use this information to assist decision-making, is the main problem solved by business intelligence (BI).
How to transform the data existing in the database into the information needed by business personnel? Most of the answers are report systems. In short, the report system can already be called BI, which is the low-end implementation of BI. At present, most foreign enterprises have entered the mid-range BI, which is the so-called data analysis. Some enterprises have begun to enter the high-end BI, which is data mining. However, most enterprises in China are still in the declaration stage. At present, there are many domestic BI manufacturers. Well-known BI vendors, such as Yonghong Technology, the core product Yonghong -Z-Suite, help enterprises build big data applications, and also have rich industry accumulation, such as government, electricity, energy and finance. If you are interested, you can check some information yourself.