Question? Leave a message!




Data Warehouses and OLAP

Data Warehouses and OLAP
Data Warehouses and OLAP www.ThesisScientist.comData Warehousing and OLAP Technology for Data Mining  What is a data warehouse  A multidimensional data model  Data warehouse architecture  Data warehouse implementation  Further development of data cube technology  From data warehousing to data mining www.ThesisScientist.comWhy Data Warehousing  Data warehousing can be considered as an important preprocessing step for data mining Heterogeneous Databases data selection Data Warehouse data cleaning data integration data summarization  A data warehouse also provides online analytical processing (OLAP) tools for interactive multidimensional data analysis. www.ThesisScientist.comExample of a Data Warehouse (1) USDatabase Data Warehouse Employee Department FACT table eid name birthdate did dname ... ... ... ... ... timeid pid sales 1 1 2 2 1 4 Transaction Details 2 2 1 tid type date tid pid qty 3 3 2 1 sale 4/11/1999 1 21 2 ... ... ... 2 sale 5/2/1999 2 13 1 3 buy 5/17/1999 3 41 3 dimension 1: time ... ... ... ... ... ... timeid day month year 1 11 4 1999 2 15 4 1999 HKDatabase 3 2 5 1999 Supplier Country ... ... ... sid name birthdate cid cname ... ... ... ... ... dimension 2: product pid name type sid date time qty pid 1 chair office 1 15:4:1999 8:30 2 11 2 table office 2 15:4:1999 9:30 2 11 Sales 3 desk office 3 3 56 ... ... 4 19:5:1999 4 22 www.ThesisScientist.com ... ...Example of a Data Warehouse (2)  Data Selection  Only data which are important for analysis are selected (e.g., information about employees, departments, etc. are not stored in the warehouse)  Therefore the data warehouse is subjectoriented  Data Integration  Consistency of attribute names  Consistency of attribute data types. (e.g., dates are converted to a consistent format)  Consistency of values (e.g., productids are converted to correspond to the same products from both sources)  Integration of data (e.g, data from both sources are integrated into the warehouse) www.ThesisScientist.comExample of a Data Warehouse (3)  Data Cleaning  Tuples which are incomplete or logically inconsistent are cleaned  Data Summarization  Values are summarized according to the desired level of analysis  For example, HK database records the daytime a sales transaction takes place, but the most detailed time unit we are interested for analysis is the day. www.ThesisScientist.comExample of a Data Warehouse (4)  Example of an OLAP query (collects counts)  Summarize all company sales according to product and year, and further aggregate on each of these dimensions. year 1999 2000 2001 2002 ALL chairs 25 37 89 21 172 tables 10 30 0 45 85 Data cube desks 56 84 9 35 184 shelves 19 20 0 71 110 boards 5 16 11 15 47 ALL 115 187 109 187 598 www.ThesisScientist.com productWhat is Data Warehouse  Defined in many different ways, but not rigorously.  A decision support database that is maintained separately from the organization‟s operational database  Support information processing by providing a solid platform of consolidated, historical data for analysis.  “A data warehouse is a subjectoriented, integrated, timevariant, and nonvolatile collection of data in support of management‟s decisionmaking process.”—W. H. Inmon  Data warehousing:  The process of constructing and using data warehouses www.ThesisScientist.comData Warehouse—SubjectOriented  Organized around major subjects, such as customer, product, sales.  Focusing on the modeling and analysis of data for decision makers, not on daily operations or transaction processing.  Provide a simple and concise view around particular subject issues by excluding data that are not useful in the decision support process. www.ThesisScientist.comData Warehouse—Integrated  Constructed by integrating multiple, heterogeneous data sources  relational databases, flat files, online transaction records  Data cleaning and data integration techniques are applied.  Ensure consistency in naming conventions, encoding structures, attribute measures, etc. among different data sources  E.g., Hotel price: currency, tax, breakfast covered, etc.  When data is moved to the warehouse, it is www.ThesisScientist.com converted. Data Warehouse—Time Variant  The time horizon for the data warehouse is significantly longer than that of operational systems.  Operational database: current value data.  Data warehouse data: provide information from a historical perspective (e.g., past 510 years)  Every key structure in the data warehouse  Contains an element of time, explicitly or implicitly  But the key of operational data may or may not contain “time element” (the time elements could be extracted from log files of transactions) www.ThesisScientist.comData Warehouse—NonVolatile  A physically separate store of data transformed from the operational environment.  Operational update of data does not occur in the data warehouse environment.  Does not require transaction processing, recovery, and concurrency control mechanisms  Requires only two operations in data accessing:  initial loading of data and access of data. www.ThesisScientist.comData Warehouse vs. Heterogeneous DBMS  Traditional heterogeneous DB integration:  Build wrappers/mediators on top of heterogeneous databases  Query driven approach  When a query is posed to a client site, a metadictionary is used to translate the query into queries appropriate for individual heterogeneous sites involved, and the results are integrated into a global answer set  Complex information filtering, compete for resources  Data warehouse: updatedriven, high performance  Information from heterogeneous sources is integrated in advance and stored in warehouses for direct query and analysis www.ThesisScientist.comData Warehouse vs. Heterogeneous DBMS  Example of a Heterogeneous DBMS Heterogeneous Databases mediator/ wrapper R1 Q1 results meta R2 user data Q2 query R3 query Q3 transformation  The results from the various sources are integrated and returned to the user www.ThesisScientist.comData Warehouse vs. Heterogeneous DBMS  Advantages of a Data Warehouse:  The information is integrated in advance, therefore there is no overhead for (i) querying the sources and (ii) combining the results  There is no interference with the processing at local sources (a local source may go offline)  Some information is already summarized in the warehouse, so query effort is reduced.  When should mediators be used  When queries apply on current data and the information is highly dynamic (changes are very frequent).  When the local sources are not collaborative. www.ThesisScientist.comData Warehouse vs. Operational DBMS  OLTP (online transaction processing)  Major task of traditional relational DBMS  Daytoday operations: purchasing, inventory, banking, manufacturing, payroll, registration, accounting, etc.  OLAP (online analytical processing)  Major task of data warehouse system  Data analysis and decision making  Distinct features (OLTP vs. OLAP):  User and system orientation: customer vs. market  Data contents: current, detailed vs. historical, consolidated  Database design: ER + application vs. star + subject  View: current, local vs. evolutionary, integrated  Access patterns: update vs. readonly but complex queries www.ThesisScientist.comOLTP vs. OLAP OLTP OLAP users clerk, IT professional manager function day to day operations Decision support DB design applicationoriented subjectoriented data current, uptodate historical, detailed, flat relational summarized, multidimensional isolated integrated, consolidated usage repetitive adhoc access read/write lots of scans index/hash on prim. key unit of work short, simple transaction complex query records accessed tens millions users thousands hundreds DB size 100MBGB 100GBTB (even PB) metric transaction throughput query throughput, response www.ThesisScientist.com Why Separate Data Warehouse  High performance for both systems  DBMS— tuned for OLTP: access methods, indexing, concurrency control, recovery  Warehouse—tuned for OLAP: complex OLAP queries, multidimensional view, consolidation.  Different functions and different data:  missing data: Decision support requires historical data which operational DBs do not typically maintain  data consolidation: DS requires consolidation (aggregation, summarization) of data from heterogeneous sources  data quality: different sources typically use inconsistent data representations, codes and formats which have to be reconciled www.ThesisScientist.comData Warehousing and OLAP Technology for Data Mining  What is a data warehouse  A multidimensional data model  Data warehouse architecture  Data warehouse implementation  Further development of data cube technology  From data warehousing to data mining www.ThesisScientist.comFrom Tables and Spreadsheets to Data Cubes  A data warehouse is based on a multidimensional data model which views data in the form of a data cube  A data cube, such as sales, allows data to be modeled and viewed in multiple dimensions  Dimension tables, such as item (itemname, brand, type), or time(day, week, month, quarter, year)  Fact table contains measures (such as dollarssold) and keys to each of the related dimension tables www.ThesisScientist.comFrom Tables and Spreadsheets to Data Cubes  A dimension is a perspective with respect to which we analyze the data  A multidimensional data model is usually organized around a central theme (e.g., sales). Numerical measures on this theme are called facts, and they are used to analyze the relationships between the dimensions  Example:  Central theme: sales  Dimensions: item, customer, time, location, supplier, www.ThesisScientist.com etc.What is a data cube  The data cube summarizes the measure with respect to a set of n dimensions and provides summarizations for all subsets of them year 1999 2000 2001 2002 ALL chairs 25 37 89 21 172 tables 10 30 0 45 85 desks Data cube 56 84 9 35 184 shelves 19 20 0 71 110 boards 5 16 11 15 47 ALL 115 187 109 187 598 www.ThesisScientist.com productWhat is a data cube  In data warehousing literature, the most detailed part of the cube is called a base cuboid. The top most 0D cuboid, which holds the highestlevel of summarization, is called the apex cuboid. The lattice of cuboids forms a data cube. year base cuboid 1999 2000 2001 2002 ALL chairs 25 37 89 21 172 tables 10 30 0 45 85 desks Data cube 56 84 9 35 184 shelves 19 20 0 71 110 apex cuboid boards 5 16 11 15 47 ALL 115 ww187 w.ThesisS 109 cientist.co 187 m 598 productCube: A Lattice of Cuboids all 0D(apex) cuboid time item location supplier 1D cuboids time,item time,location item,location location,supplier 2D cuboids time,supplier item,supplier time,location,supplier time,item,location 3D cuboids time,item,supplier item,location,supplier 4D(base) cuboid www.ThesisScientist.com time, item, location, supplierConceptual Modeling of Data Warehouses  The ER model is used for relational database design. For data warehouse design we need a concise, subjectoriented schema that facilitates data analysis.  Modeling data warehouses: dimensions measures  Star schema: A fact table in the middle connected to a set of dimension tables  Snowflake schema: A refinement of star schema where some dimensional hierarchy is normalized into a set of smaller dimension tables, forming a shape similar to snowflake  Fact constellations: Multiple fact tables share dimension tables, viewed as a collection of stars, therefore called galaxy schema or fact constellation www.ThesisScientist.comExample of Star Schema time item timekey foreign keys day itemkey dayoftheweek Sales Fact Table itemname month brand timekey quarter type year suppliertype itemkey branchkey location branch locationkey locationkey branchkey street branchname unitssold city branchtype dollarssold provinceorstreet country avgsales Measures www.ThesisScientist.comExample of Snowflake Schema time item timekey itemkey day supplier Sales Fact Table dayoftheweek itemname supplierkey brand month suppliertype timekey type quarter supplierkey year itemkey branchkey location branch locationkey locationkey branchkey street unitssold branchname citykey city branchtype dollarssold citykey city avgsales provinceorstreet Measures normalization country www.ThesisScientist.comExample of Fact Constellation time item Shipping Fact Table timekey day itemkey dayoftheweek timekey Sales Fact Table itemname month brand itemkey quarter type timekey year suppliertype shipperkey itemkey fromlocation branchkey tolocation branch locationkey location branchkey dollarscost locationkey unitssold branchname street unitsshipped branchtype city dollarssold provinceorstreet avgsales country shipper Measures shipperkey shippername www.ThesisScientist.com locationkey shippertypeA Data Mining Query Language, DMQL: Language Primitives  Cube Definition (Fact Table) define cube cubename dimensionlist: measurelist  Dimension Definition ( Dimension Table ) define dimension dimensionname as (attributeorsubdimensionlist)  Special Case (Shared Dimension Tables)  First time as “cube definition”  define dimension dimensionname as dimensionnamefirsttime in cube cubenamefirsttime www.ThesisScientist.comDefining a Star Schema in DMQL define cube salesstar time, item, branch, location: dollarssold = sum(salesindollars), avgsales = avg(salesindollars), unitssold = count() define dimension time as (timekey, day, dayofweek, month, quarter, year) define dimension item as (itemkey, itemname, brand, type, suppliertype) define dimension branch as (branchkey, branchname, branchtype) define dimension location as (locationkey, street, city, provinceorstate, country) www.ThesisScientist.comDefining a Snowflake Schema in DMQL define cube salessnowflake time, item, branch, location: dollarssold = sum(salesindollars), avgsales = avg(salesindollars), unitssold = count() define dimension time as (timekey, day, dayofweek, month, quarter, year) define dimension item as (itemkey, itemname, brand, type, supplier(supplierkey, suppliertype)) define dimension branch as (branchkey, branchname, branchtype) define dimension location as (locationkey, street, www.ThesisScientist.com city(citykey, provinceorstate, country))Defining a Fact Constellation in DMQL define cube sales time, item, branch, location: dollarssold = sum(salesindollars), avgsales = avg(salesindollars), unitssold = count() define dimension time as (timekey, day, dayofweek, month, quarter, year) define dimension item as (itemkey, itemname, brand, type, suppliertype) define dimension branch as (branchkey, branchname, branchtype) define dimension location as (locationkey, street, city, provinceorstate, country) define cube shipping time, item, shipper, fromlocation, tolocation: dollarcost = sum(costindollars), unitshipped = count() define dimension time as time in cube sales define dimension item as item in cube sales define dimension shipper as (shipperkey, shippername, location as location in cube sales, shippertype) define dimension fromlocation as location in cube sales define dimension tolocation as location in cube sales www.ThesisScientist.comAggregate Functions on Measures: Three Categories  distributive: if the result derived by applying the function to n aggregate values is the same as that derived by applying the function on all the data without partitioning.  E.g., count(), sum(), min(), max().  algebraic: if it can be computed by an algebraic function with M arguments (where M is a bounded integer), each of which is obtained by applying a distributive aggregate function.  E.g., avg(), minN(), standarddeviation().  holistic: if there is no constant bound on the storage size needed to describe a subaggregate. www.ThesisScientist.com  E.g., median(), mode(), rank().Aggregate Functions on Measures: Three Categories (Examples)  Table: Sales(itemid, timeid, quantity)  Target: compute an aggregate on quantity  distributive:  To compute sum(quantity) we can first compute sum(quantity) for each item and then add these numbers.  algebraic:  To compute avg(quantity) we can first compute sum(quantity) and count(quantity) and then divide these numbers.  holistic:  To compute median(quantity) we can use neither median(quantity) for each item nor any combination of distributive functions, too. www.ThesisScientist.comConcept Hierarchies  A concept hierarchy is a hierarchy of conceptual relationships for a specific dimension, mapping lowlevel concepts to highlevel concepts  Typically, a multidimensional view of the summarized data has one concept from the hierarchy for each selected dimension  Example:  General concept: Analyze the total sales with respect to item, location, and time  View 1: itemid, city, month  View 2: itemtype, country, week  View 3: itemcolor, state, year www.ThesisScientist.com  ....A Concept Hierarchy: Dimension (location) all all Europe ... NorthAmerica region Germany ... Spain Canada ... Mexico country Vancouver ... city Frankfurt ... Toronto L. Chan ... M. Wind office www.ThesisScientist.comView of Warehouses and Hierarchies Specification of hierarchies  Schema hierarchy day month quarter; week year  Setgrouping hierarchy 1..10 inexpensive www.ThesisScientist.comMultidimensional Data  Sales volume as a function of product, month, and region Dimensions: Product, Location, Time Hierarchical summarization paths Industry Region Year Category Country Quarter Product City Month Week Office Day total order partial order Month www.ThesisScientist.com (lattice) ProductA Sample Data Cube Total annual sales Date of TV in U.S.A. 2Qtr 1Qtr 3Qtr sum 4Qtr TV U.S.A PC VCR sum Canada Mexico sum www.ThesisScientist.com CountryCuboids Corresponding to the Cube all 0D(apex) cuboid country product date 1D cuboids product,date product,country date, country 2D cuboids 3D(base) cuboid product, date, country The cuboids are also called multidimensional views www.ThesisScientist.comDataCube example „color‟, „size‟: DIMENSIONS „count‟: MEASURE C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comDataCubes „color‟, „size‟: DIMENSIONS „count‟: MEASURE C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comDataCubes „color‟, „size‟: DIMENSIONS „count‟: MEASURE C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comDataCubes „color‟, „size‟: DIMENSIONS „count‟: MEASURE C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comDataCubes „color‟, „size‟: DIMENSIONS „count‟: MEASURE C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comDataCubes „color‟, „size‟: DIMENSIONS „count‟: MEASURE C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size DataCube www.ThesisScientist.comBrowsing a Data Cube  Visualization  OLAP capabilities  Interactive manipulation www.ThesisScientist.comTypical OLAP Operations  Browsing between cuboids  Roll up (drillup): summarize data  by climbing up hierarchy or by reducing a dimension  Drill down (roll down): reverse of rollup  from higher level summary to lower level summary or detailed data, or introducing new dimensions  Slice and dice:  project and select  Pivot (rotate):  reorient the cube, visualization, 3D to series of 2D planes.  Other operations  drill across: involving (across) more than one fact table  drill through: through the bottom level of the cube to its backend relational tables (using SQL) www.ThesisScientist.comExample of operations on a Datacube C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comExample of operations on a Datacube Rollup:  In this example we reduce one dimension  It is possible to climb up one hierarchy  Example (product, city)  (product, country) C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comExample of operations on a Datacube Drilldown  In this example we add one dimension  It is possible to climb down one hierarchy  Example (product, year)  (product, month) C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comExample of operations on a Datacube Slice: Perform a selection on one dimension C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comExample of operations on a Datacube Dice: Perform a selection on two or more dimensions C / S S M L TOT f Red 20 3 5 28 size Blue 3 3 8 14 color Gray 0 0 5 5 TOT 23 6 18 47 color; size www.ThesisScientist.comA StarNet Query Model (contracts,group,district,country,qtrly) Customer Orders Shipping Method Customer CONTRACTS AIREXPRESS ORDER TRUCK PRODUCT LINE Time Product ANNUALY QTRLY DAILY PRODUCT ITEM PRODUCT GROUP CITY SALES PERSON COUNTRY DISTRICT REGION DIVISION Each circle is Location www.ThesisScientist.com Promotion Organization called a footprintData Warehousing and OLAP Technology for Data Mining  What is a data warehouse  A multidimensional data model  Data warehouse architecture  Data warehouse implementation  Further development of data cube technology  From data warehousing to data mining www.ThesisScientist.comDesign of a Data Warehouse: A Business Analysis Framework  Four views regarding the design of a data warehouse  Topdown view  allows selection of the relevant information necessary for the data warehouse  Data source view  exposes the information being captured, stored, and managed by operational systems  Data warehouse view  consists of fact tables and dimension tables  Business query view  sees the perspectives of data in the warehouse from the view of enduser www.ThesisScientist.comData Warehouse Design Process  Topdown, bottomup approaches or a combination of both  Topdown: Starts with overall design and planning  Bottomup: Starts with experiments and prototypes (rapid)  From software engineering point of view  Waterfall: structured and systematic analysis at each step before proceeding to the next  Spiral: rapid generation of increasingly functional systems, short turn around time, quick turn around  Typical data warehouse design process  Choose a business process to model, e.g., orders, invoices, etc.  Choose the grain (atomic level of data) of the business process  Choose the dimensions that will apply to each fact table record  Choose the measure that will populate each fact table record www.ThesisScientist.comMultiTiered Architecture Monitor OLAP Server Metadata other Integrator sources Analysis Extract Query Operational Serve Transform Data Reports DBs Load Warehouse Data mining Refresh Data Marts www.ThesisScientist.com Data Sources Data Storage OLAP Engine FrontEnd ToolsThree Data Warehouse Models  Enterprise warehouse  collects all of the information about subjects spanning the entire organization  Data Mart  a subset of corporatewide data that is of value to a specific groups of users. Its scope is confined to specific, selected groups, such as marketing data mart  Independent vs. dependent (directly from warehouse) data mart  Virtual warehouse  A set of views over operational databases  Only some of the possible summary views may be materialized www.ThesisScientist.comDevelopment: A Recommended Approach MultiTier Data Warehouse Distributed Data Marts Enterprise Data Data Data Mart Mart Warehouse Model refinement Model refinement Define a highlevel corporate data model www.ThesisScientist.com
Website URL
Comment