Typically, a database trading
consists of a file in which each record represents a transaction. Generally, a transaction consists of a unique transaction identification number, and a list of items of transaction (e.g., purchase of goods in the shop). Transaction database may have some additional tables associated with it, contains information about the sales of other, such as the date of the transaction, the customer ID number, the seller's ID number, sales outlets, and so on.
If we want to dig deeper into the data, in the business operation, ask "what goods are suitable to sell together?" This "basket data analysis" enables us to bundle items into groups as a strategy for expanding sales. For example, given the knowledge that printers and computers are often sold together, you can offer customers who buy a selected computer a discount on an expensive printer and hope to sell more expensive printers. Conventional data retrieval systems cannot answer such queries. However, a data mining system for transactional data can do this by identifying items that are frequently sold together. Here we mainly study the statistical methods of data mining for transactional databases.
Differences between database transactions and database analysis:
are mainly real-time, application-oriented database, response timeliness requirements are very high, only focus on the most recent period of data. Analytical database is mainly used to analyze laws in a large amount of data. Generally, the stored data span is long, the data volume is large, and the real-time requirement is not high. Through the inquiry to analysis rule tendency, it can be used in the product decision-making and so on.