SharePoint 2010 Business Intelligence Module 6: Analysis Services.
Post on 26-Dec-2015
214 Views
Preview:
Transcript
SharePoint 2010 Business Intelligence
Module 6: Analysis Services
Overview
Analysis Services
Lesson: Analysis Services
Introduction
ETL
OLAP Terms
Storage Modes
Queries
Tools
Mining Models
Introduction
Analysis Services provides access to large data sets
Running SQL queries against a 100 million row table just doesn’t work
When your data sets get large, you need a better way of handling the load
Online analytical processing (OLAP) provides all your answers with speed!
Analysis services is an OLAP implementation
ETL
The Extract, Transform and Load (ETL) process is vital to the OLAP results
If you input junk, you get junk back out! All data that enters the OLAP database must be valid or
your results could be exponentially wrong!Make sure that all possible means are employed when ensuring only valid data is entered into the system and that it is entered only once!
Common issue is data inserted more than once which causes invalid results
OLAP Terms
Data Source
A source or destination of data Fact table
A table that contains numerical information with keys that map to the dimension keysMeasure
A single numerical value in a fact tableDimension
A set of labels/attributes that describe the measures in a fact tableCube
A set of aggregations of all the dimensions and facts complied together to produce valuable information
Perspective
A subset of dimensions and measures specific to some group of usersData Modeling
Using statistical analysis to determine patterns in large sets of data
Tools
Visual Studio / BI Workbench
Several project templates to facilitate the creation of cubes, dimensions and work with data sources/views
IntelliCube
A heuristic analysis tool for automatic generation of a cube based on its data and relationships
External Viewer forOutliersCandidate keysValue distributionsPatterns
Business Intelligence Development Studio
BI Studio is just Visual Studio with project templates installed
Project templates provide item templates and wizards
New 2008 Wizards are much easier to use and more powerful
Context sensitive functionality keeps toolbars and menus trimmed to what elements you need based on what you are looking at
Because it is Visual Studio it is fully customizable
3rd party add-ons can make it even more powerful
Designing an Analysis Services Database
BI Studio will be used to create new AS Databases
Steps include:
Define data source Define data view Create a new cube Define fact tables Define dimensions (data and time) Define measures Aggregate/Run the cube
Key Performance Indicators (KPI)
KPIs are used to show very simply whether a target is being accomplished
Components include the Goal, Value, Status, and Trend Example: Sales and Quality targets
Analysis Services allows you to build MDX expressions off of Cube data to build KPIs
KPI values can be queried from client applications Allows for visual display of meaningful data
Actions
Actions
Allows client application users to be able to interact with what the data means
Example: Browse to a customer or product via URLTypes of Actions:
CommandLine, DataSet, Drillthrough, Html, Custom, Report, URL
Define and assign to objects in the Cube
Perspectives
Perspectives work similar to views in a relational database
Shows different users the data they need to see for a particular role they may be in
Used to reduce complexity of cube data
Not meant to be security mechanism for data
Storage Modes
Data in an Analysis Services database is stored differently than a relational database
Optimized storage provides the OLAP query performancePartition
MOLAP – multidimensional OLAP (fact data and aggregations are stored in special format)
ROLAP – Relational OLAP (fact data and aggregations remain in relational database)
HOLAP – Hybrid OLAP (fact data is relational, aggregations are stored in special format)
Dimension (dimension attributes only)
MOLAP – stored in special format ROLAP – stay in relational format
Querying Cubes
SQL is not used in OLAP databases
MDX (multidimensional queries) is used for querying cubes
DMX (Data Mining queries) is used for querying data mining models
MDX Queries
Multidimensional Expressions (MDX) are used to query multidimensional data
Some common terms are:
Cell – the space at an intersection of a measure and attribute
Tuple – a unique cell based on a set of attribute members Set – an order set of tuples with same dimensionality
Calculated Members and Named Sets
Calculated Members are used when you need to determine something at query time
Can be query or session scoped Value are only stored in memory not on disk
Named Sets are basically predefined MDX queries that can be reused in other queries
Used to group dimension members
Analyzing Data with Data Mining Algorithms
Out of the box, Analysis services provides five algorithms:
Classification (Decision Tree)Predict one or more discrete variables, based on the other attributes in the dataset
Regression (Time Series)Predict one or more continuous variables, such as profit or loss, based on other attributes in the dataset
Segmentation (Clustering)Divide data into groups, or clusters, of items that have similar properties.
Association (Association)Find correlations between different attributes in a dataset
Sequence analysis (Sequence Clustering)Summarize frequent sequences or episodes in data, such as a Web path flow
What could go wrong?
Cubes really are simple things to build and utilize
As simple as they are, it is easy to create a cube that has the wrong data
Always validate that the data that is displayed in the Cube is valid and accurate
Never hurts to be overly aggressive when testing Cube data
Lab 1: Analysis Services
Explore Analysis Services
Lab 2: Building a Cube
Explore Cubes and Dimensions
Lab 3: Data Mining Algorithms
Explore Data Mining Models
Review
Your instructor will ask a series of questions on this module
Summary
Extra Large databases are not easily queried for data
Analysis Services is an OLAP tool to manage large databases
Ensure that your ETL process is accurate
Data Mining Algorithms can help you find patterns you didn’t know about before
top related