Understanding Dimensional Models

转载 2011年01月24日 16:38:00

 


To analyze business data, the data needs to be mapped logically to a business model. The Siebel Analytics Server can use dimensional models for this purpose. This section discusses some of the components and variants of representative dimensional models.

usinesses are analyzed by relevant dimensional criteria, and the business model is developed from these relevant dimensions. These dimensional models form the basis of the valid business models to use with the Siebel Analytics Server. All dimensional models build on a star schema. That is, they model some measurable facts that are viewed in terms of various dimensional attributes.

imensional Hierarchies

Dimensions are categories of attributes by which the business is defined. Common dimensions are time periods, products, markets, customers, suppliers, promotion conditions, raw materials, manufacturing plants, transportation methods, media types, and time of day. Within a given dimension, there may be many attributes. For example, the time period dimension can contain the attributes day, week, month, quarter, and year. Exactly what attributes a dimension contains depends on the way the business is analyzed.

A dimensional hierarchy expresses the one-to-many relationships between attributes. Given a sample time dimension, consider the hierarchies it implies, as shown in Figure 6 .

Figure 6.  Sample Hierarchy

image

With this sample time dimension, days may aggregate, or roll up, into weeks. Months may roll up into quarters, and quarters into years. When one attribute rolls up to another, it implies a one-to-many relationship. The total of all the hierarchy definitions in this sample time dimension make up this time dimension.

These hierarchy definitions have to be specific to the business model—one model may be set up where weeks roll up into a year, and another where they do not. For example, in a model where weeks roll up into a year, it is implied that each week has exactly one year associated with it; this may not hold true for calendar weeks, where the same week could span two years.

Some hierarchies might require multiple elements to roll up, as when the combination of month and year roll up into exactly one quarter. The Siebel Analytics Server allows you to define the hierarchy definitions for your particular business, however complex, assuring that analyses will conform to your business definitions.

Factual Measures

Factual measures, or facts, are typically additive data such as dollar value or quantity sold, and they can be specified in terms of the dimensions. For example, you might ask for the sum of dollars for a given product in a given market over a given time period. Measures can also be aggregated by applying other aggregation rules, such as averaging instead of summing. Furthermore, the aggregation rules can be specific to particular dimensions. The Siebel Analytics Server allows you to define these complex, dimension-specific aggregation rules.

Star and Snowflake Models

Star and snowflake models follow the dimensional rules of one-to-many relationships. Star schemas have one-to-many relationships between the logical dimension tables and the logical fact table. Snowflake schemas have those same types of relationships, but also include one-to-many relationships between elements in the dimensions.

Bridge Tables to Model Many-to-Many Relationships

Star schemas and snowflake schemas work well for modeling a particular part of a business where there are one-to-many relationships between the dimension tables and the fact tables. However, sometimes it is necessary to model many-to-many relationships between dimension tables and fact tables.

When you need to model many-to-many relationships between dimension tables and fact tables, you can create a bridge table that resides between the fact table and the dimension table. A bridge table stores multiple records corresponding to that dimension.

To understand how a bridge table works, consider the following portion of a sample health care schema, as shown in Figure 7 .

Figure 7.  Sample Health Care Schema

image

The many-to-many relationship is that for each patient admission, there can be multiple diagnoses. For example, a patient can be diagnosed with the flu and with a broken wrist. The bridge table then needs to have a weight factor column in it so that all of the diagnoses for a single admission add up to a value of 1. The weight factor has to be calculated as part of the process of building the data. For the case of the patient diagnosed with the flu and a broken wrist, there would be one record in the Admission Records table, two records in the Diagnosis Record table, and two records in the Diagnosis table, as shown in Figure 8 .

Figure 8.  Multiple Diagnoses for One Patient

image

NOTE: This type of design can create more records in the Diagnosis Records table than in the Admission Records table. You can limit the number of records in the Diagnosis Records table by predefining groups of diagnosis and forcing each admission record to fit in one of these predefined groups.

In the Administration Tool, the Logical Table dialog box has an option you can select to specify that a table is a bridge table.

Single Table Models

For the greatest simplicity for end users, you can construct a subject area that consists of a single table. To create a single table model, you first create a logical dimensional model, and then present it as a single table schema in the Administration Tool's Presentation layer. The logical dimensional model is required to set up the necessary metadata for the Siebel Analytics Server to navigate to the proper physical tables. For information about the Presentation layer, see Creating and Maintaining the Presentation Layer in a Repository .

一份关于深度学习论文的总结笔记

原文地址:http://www.tuicool.com/articles/miQZBzn NLP Strategies for Training Large Vocabulary Neur...
  • wanzew
  • wanzew
  • 2016年10月30日 19:40
  • 851

Topic Model 相关论文

ICML 2014 1 、Jian Tang, ZhaoshiMeng, XuanLongNguyen, Qiaozhu Mei, MingZhang: Understanding the L...
  • liujinwen1992
  • liujinwen1992
  • 2014年07月02日 21:13
  • 982

《深入理解Java虚拟机》--Understanding the Jvm(上)

《深入理解Java虚拟机》–Understanding the Jvm(上)前言:跟”Thinking in Java”不同的是,《深入理解Java虚拟机》是一本修炼内功心法的书。因为虚拟机对开发者来...
  • change_on
  • change_on
  • 2016年10月07日 12:30
  • 905

关于freecodecamp网站

自学前端有半个月了,基本了解了HTML、CSS、JS、Jquery的知识,又在慕课网上巩固了HTML+CSS。昨天自己用最原始的方法做了了静态的百度首页,虽然耗时一天,但是从中发现了所学知识在实践中会...
  • catherine0923
  • catherine0923
  • 2016年09月01日 17:13
  • 153

《UNDERSTANDING COMICS》——产品经理必读

小时候比较喜欢看漫画书,而这是一本关于漫画的漫画,读了之后,才发现这是一本关于认知心理学的”正儿八经”的书。 我之所以在标题中加入“产品经理必读”字样,是因为: 我做标题党了,也许这样...
  • zhengjie19
  • zhengjie19
  • 2015年09月08日 10:40
  • 322

[深度学习论文笔记][Visualizing] 网络可视化部分论文导读

There are several ways to understanding and visualing CNN 1 Visualizing Activations Show the a...
  • Hao_Zhang_Vision
  • Hao_Zhang_Vision
  • 2016年10月29日 10:19
  • 461

卷积神经网络可视化套路

类型一:公式类型 此种类型,无非是建立一个可视化模型,公式有两部分组成,前半部分是主要部分,后半部分是正则项或类似正则项的东西。 《Understanding Neural Networks T...
  • u011204487
  • u011204487
  • 2016年06月23日 16:17
  • 897

django models进行数据库增删查改

引入models的定义  from app.models import  myclass  class  myclass():       aa =  models. CharField (ma...
  • theowl
  • theowl
  • 2015年09月23日 15:03
  • 1141

django models进行数据库增删查改

引入models的定义 from app.models import  myclass class  myclass():      aa =  models. CharField (max_l...
  • u014368609
  • u014368609
  • 2016年06月13日 09:42
  • 769

了解Nginx HTTP代理、负载均衡、缓冲和缓存

提供:ZStack云计算 前言本文将讨论Nginx的HTTP代理功能,该功能可以将前端请求转发到后端服务器。规模较大的应用经常会使用Nginx作为反向代理,以处理后端服务器处理不了的请求量。我们还将讨...
  • zstack_org
  • zstack_org
  • 2016年12月30日 10:25
  • 2883
内容举报
返回顶部
收藏助手
不良信息举报
您举报文章:Understanding Dimensional Models
举报原因:
原因补充:

(最多只允许输入30个字)