Data Warehouse

Knowledge Discovery Process

OLTP & OLAP

联机事务处理（OLTP, online transactional processing）系统：涵盖组织机构大部分的日常操作，purchasing, inventory, banking,manufacturing, payroll, registration, accounting

联机分析处理（OLAP, online analytical processing）系统：以不同的格式组织和提供数据，以满足不同用户的各种需求，为数据分析和决策方面提供服务。

 

Distinct features (OLTP vs. OLAP):

 User and system orientation: customer vs. market

 Data contents: current, detailed vs. historical, consolidated

 View: current, local vs. evolutionary, integrated

 Access patterns: update vs. read-only but complex queries

Data Warehouse

DBMS— tuned for OLTP: access methods, indexing, concurrency control, recovery

Warehouse—tuned for OLAP: complex OLAP queries, multidimensional view, consolidation

 

Data Warehouse:

 

数据仓库将分布在企业网络中不同信息岛上的业务数据集成到一起，存储在一个单一的集成关系型数据库中，利用这样的集成信息，可方便用户对信息访问，可使决策人员对一段时间内的历史数据进行分析，研究事务的发展走势。

 

A data warehouse is a subject-oriented, integrated, time-variant, and nonvolatile collection of data in support of management’s decision-making process.” — W. H.Inmon

 

data stored in data warehouse has been processed after extracation, cleaning, transformation, load(sort, summarize...) and refresh.

 

 

 

Data Warehouse model : dimensions and measures, you can locate some data by dimension and see the data by measures 

Conception model : star schema, snowflake schema(a refinement of star schema), fact constellations(a collection of stars)

Example of Star Schema:

 

Typical OLAP Operations : 

Roll up: summarize data by climbing up hierarchy or by dimension reduction, you can roll up to all to reduce a dimension

Dill down: reverse of Roll-up, from higher level summary to lower level summary or detailed data

Slice and dice: project and select 

Priot(rotate): reorient the cube, visualization, 3D to series of 2D planes.

 

参考

中国科学院大学《数据挖掘》课程slices