Business Intelligence
Combination of technologies like
- Data Warehousing (DW)
- On-Line Analytical Processing (OLAP)
- Data Mining (DM)
- Data Visualization (VIS)
- Decision Analysis (what-if)
- Customer Relationship Management (CRM)
Operational Data
- Presents a dynamic view of the business
- Must be kept up-to-date and current at all times
- Updated by transactions entered by data-entry operators or specially trained end users Is maintained in detail
- Utilization is predictable. Systems can be optimized for projected workloads
- High volume of transactions, each of which affects a small portion of the data
- Users do not need to understand data structures
- Functional orientation
Analytical Data
- Presents a static view of the business
- End-user access is usually read-only
- More concerned with summary information
- Usage is unpredictable in terms of depth of information needed by the user
- Smaller number of queries, each of which may access large amounts of data
- Users need to understand the structure of the data (and business rules) to draw meaningful conclusions from the data
- Subject -orientation
Database
Broadly classified into
- OLTP (Online Transactional Processing) DB
- OLAP (Online Analytical Processing) DB
OLAP
- Slicing and dicing of data is called as Online Analytical Processing (OLAP). OLAP only serves the needs of data warehousing than OLTP.
- OLAP systems allow ad hoc processing and support access to data over time periods.
- OLAP systems are the aggregation, transformation, integration and historical collection of OLTP data from one or more systems.
- Typical OLAP operations:
- Roll up (drill up)
- summarize data by climbing hierarchy or by dimension reduction.
- Drill down(roll down)
- 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.
OLAP vs OLTP
Slno
|
OLTP
|
OLAP
|
1.
|
Transaction Oriented
|
Decision Oriented (Reports)
|
2.
|
Complex data model (fully normalized)
|
Simple data model
(multidimensional/de-normalized)
|
3.
|
Smaller data volume (few historical data)
|
Larger data volumes (collection of historical data)
|
4.
|
Many, ”small” queries
|
Fewer, but ”bigger” queries
|
5.
|
Frequent updates
|
Frequent reads, in-frequent updates (daily)
|
6.
|
Huge no. of users(clerks).
|
Only few users(Management Personnel)
|
Objective of Data Warehouse
The primary purpose of a data warehouse is to provide easy access to specially prepared data that can be used with decision support applications, such as management reporting, queries, decision support systems, and executive information systems.
Decision Support
A Decision Support System (DSS) is a system that provides managers with information they need to make decisions. These systems have the effect of empowering employees at all levels, providing them access to business and financial information that directly impact their productivity and quality of work
Executive information systems
An Executive information system (EIS) is a concise snapshot of how the company is doing today. Consider it as an electronic executive briefing. EIS allows greater flexibility in “slicing-and-dicing” data, i.e.; it allows exploration of data through multiple dimensions or views.
Why Datawarehouse?
By centralizing data
- The queries can be answered locally without accessing the original information sources. Thus, high query performance can be obtained for complex aggregation queries that are needed for in-depth analysis, decision support and data mining – a way of extracting relevant data from a vast database.
- On-line Analytical Processing (OLAP) is decoupled (separated) as much as possible from On-line Transaction Processing (OLTP). Thus making information accessible to decision makers avoiding interference of OLAP with local processing at the operational sources.
Data warehouse
- A decision support database that is maintained separately from the organization’s operational databases
- A Data Warehouse is an enterprise-wise collection of
- Subject oriented
- Integrated
- Time variant
- Non-volatile
data in support of management’s decision making process.
- W. H. Inmon, 1993
*Subject Oriented - Data warehouses focuses on high-level business entities like sales,marketing,etc.
*Integrated - Data in the warehouse is obtained from multiple sources and kept in a consistent format.
*Time-Varying - Every data component in the date warehouse associates itself with some point of time like weekly,monthly,quarterly, yearly
*Non-volatile - Dw stores historical data. Data does not change once it gets into the warehouse. Only load/refresh.
Data from the operational systems are
- Extracted
- Cleansed
- Transformed
- case conversion,
- data trimming,
- concatenation,
- datatype conversion
- Aggregated
- Loaded into DW
- Periodically refreshed to reflect updates at the sources and purged from the warehouse onto slower archival storage.
Use of DWH
- Ad-hoc analyses and reports
- Data mining: identification of trends
- Management Information Systems
Designing a database for a Data Warehouse
- Define User requirements, considering different views of users from different departments.
- Identify data integrity, synchronization and security issues/bottlenecks.
- Identify technology, performance, availability & utilization requirements.
- Review normalized view of relational data to identify entities.
- Identify dimensions.
- Create and organize hierarchies of dimensions.
- Identify attributes of dimensions.
- Identify fact table(s).
- Create data repository (metadata).
10.Add calculations.
Datamart
- Datamart is a subset of data warehouse and it is designed for a particular line of business, such as sales, marketing, or finance.
- In a dependent data mart, data can be derived from an enterprise-wide data warehouse.
- In an independent data mart, data can be collected directly from sources
- May be structured for specific access tools Datamart is the data warehouse you really use Why Datamart?
- Datawarehouse projects are very expensive and time taking.
- Success rate of DWH projects is very less
To avoid single point of loss we identify department wise needs and build Datamart. If succeeded we go for other departments and integrate all datamarts into a Datawarehouse.
- Advantages
- Improve data access performance
- Simplify end-user data structures
- Facilitate ad hoc reporting
Slno
|
Data warehouse
|
Data mart
| |
1.
|
DW Operates on an enterprise level and contains all data used for reporting and analysis
|
Data Mart is used by a specific business department and is focused on a specific subject (business area).
| |
DM is a subset of DWH
|
DWH ARCHITECHTURE
Data warehouse architecture is a way of representing the overall structure of data, communication, processing and presentation that is planned, for end-user computing within the enterprise. The architecture has the following main parts:
- Operational data base
- Information access layer
- Data Access layer
- Data dictionary (metadata) layer
- Process management layer
- Application messaging layer
- Processing (Data Warehouse) layer
- Data Staging layer
Operational data is the information related to day-to-day functioning of an organization. An operational database stores business transactions critical to the functioning of the organization.
Information access layer is the layer that the end-user deals with directly. Examples of these are ad-hoc query tools like Business Objects, Power Play and Impromptu.
Data access layer is the data interchange layer. This layer provides interface between operational data bases and information access layers. The common data language used is ‘SQL’. A familiar example of a data access layer is ‘ODBC’.
Metadata layer holds a repository of Metadata information. Metadata is defined as data about data, resulting in an intelligent, efficient way to manage data. Metadata provides the structure and content of the data warehouse, source and mapping information, transformation / integration description and business rules. It is essential for quality improvement in a Data Warehouse.
Process management layer is involved in scheduling the various tasks that must be executed to build and maintain the data warehouse and data repository. It also helps to keep the Data Warehouse up-to-date.
Application messaging layer transports information around the enterprises’ computing network. It also acts as ‘middle-ware’ and isolates applications from exact data format on either end.
Processing (data warehouse) layer is the logical view of the informational data. It also performs the summarization, loading and processing of data from operational databases.
Data staging layer manages data replication across servers. It also manages data transformation.
ETL
- ETL means Extraction, transformation, and loading.
- ETL refers to the methods involved in accessing and manipulating source data and loading it into target database.
ETL Process
Etl is a process that involves the following tasks:
- extracting data from source operational or archive systems which are the primary source of data for the data warehouse
- transforming the data - which may involve cleaning, filtering, validating and applying business rules
- loading the data into a data warehouse or any other database or application that houses data
Transform
- Denormalize data
- Data cleaning.
- Case conversion
- Data trimming
- String concatenation
- datatype conversion
- Decoding
- calculation 9. Data correction.
Cleansing
The process of resolving inconsistencies and fixing the anomalies in source data, typically as part of the ETL process.
Data Staging Area
- Most complex part in the architecture.
- A place where data is processed before entering the warehouse
- It involves...
- Extraction (E)
- Transformation (T)
- Load (L)
- Indexing
Popular ETL Tools
Tool Name
|
Company Name
|
Informatica
|
Informatica Corporation
|
DT/Studio
|
Embarcadero Technologies
|
DataStage
|
IBM
|
Ab Initio
|
Ab Initio Software Corporation
|
Data Junction
|
Pervasive Software
|
Oracle Warehouse Builder
|
Oracle Corporation
|
Microsoft SQL Server Integration
|
Microsoft
|
TransformOnDemand
|
Solonde
|
Transformation Manager
|
ETL Solutions
|
Dimensional Modeling
- Means storing data in fact and dimension tables.
- Here data is fully denormalized
Dimension table
- Dimension table gives the descriptive attributes of a business.
- They are fully denormalized
- It has a primary key
- Data arranged in hierarchical manner (product to category; month to year) – if so we can use for drill down and drill up analysis
- Has less no. of records
- Has rich no. of columns
- Heavily indexed
- Dimension tables are sometimes called lookup or reference tables.
Types of Dimensions
- Normal Dimension
- Confirmed Dimension
- Junk Dimension
- Degenerated Dimension
- Role Playing Dimension
Confirmed Dimension
Dimension table used by more than one fact table is called Confirmed Dimensions
(dimensions that are linked to multiple fact tables)
D1 D2 D1 D2 D5
D3
Adv:
- To avoid unnecessary space
- Reduce time
- Drill across fact table
Junk Dimension
is an abstract dimension it will remove number of foreign keys from fact table. This is achieved by combining 2 or more dimensions into a single dimension.
Degenerated Dimension
Means a key value or dimension table which does not have descriptive attributes. i.e.) a non foreign key and non numerical measure column used for grouping purpose
Ex : Invoice Number, Ticket Number
Role Playing Dimension
Means a single physical dimension table plays different role with the help of views.
Fact Table
1. The centralized table in a star schema is called as FACT table 2. A fact table typically has two types of columns:
- Numerical measures and
- Foreign keys to dimension tables.
- The primary key of a fact table is usually a composite key that is made up of all of its foreign keys
- Fact tables store different types of measures like
- additive,
- non additive and
- semi additive measures
- A fact table might contain either detail level facts or facts that have been aggregated
- A fact table usually contains facts with the same level of aggregation.
- Has millions of records
Measure Types
- Additive - Measures that can be summarized across all dimensions. o Ex: sales
- Non Additive - Measures that cannot be summarized across all dimensions. o Ex: averages
- Semi Additive - Measures that can be summarized across few dimensions and not with others.
o Ex: inventory levels
Factless Fact
A fact table that contains no measures or facts is called as Factless Fact table.
Slowly Changing Dimensions
- Dimensions that change over time are called Slowly Changing Dimensions
- Slowly Changing Dimensions are often categorized into three types namely
- Type1,
- Type2 and
- Type3
Type 1 SCD :
- Used if history is not required Overwriting the old values.
Product Price in 2004:
Product ID(PK)
|
Year
|
Product Name
|
Product Price
|
1
|
2004
|
Product1
|
$150
|
Product Price in 2005:
Product ID(PK)
|
Year
|
Product Name
|
Product Price
|
1
|
2005
|
Product1
|
$250
|
Type 2 SCD:
- If history and current value needed
- Creating another additional record.(new record with new changes and new surrogate key)
- Mostly preferred in dimensional modeling
Product
Product ID(PK)
|
Effective
DateTime(PK)
|
Year
|
Product
Name
|
Product Price
|
Expiry DateTime
|
1
|
01-01-2004
12.00AM
|
2004
|
Product1
|
$150
|
12-31-2004
11.59PM
|
1
|
01-01-2005
12.00AM
|
2005
|
Product1
|
$250
|
Type 3 SCD:
- Used if changes are very less Previous one level of history available Creating new fields. Product Price in 2005
Product ID(PK)
|
Current Year
|
Product Name
|
Current Product Price
|
Old Product Price
|
Old Year
|
1
|
2005
|
Product1
|
$250
|
$150
|
2004
|
Surrogate keys
- Surrogate keys are always numeric and unique on a table level which makes it easy to distinguish and track values changed over time.
- Surrogate keys are integers that are assigned sequentially as needed to populate a dimension.
- Surrogate keys merely serve to join dimensional tables to the fact table.
- Surrogate keys are beneficial as the following reasons:
- Reduces space used by fact table
- Faster retrieval of data ( since alphanumerical retrieval is costlier than numerical data)
- Maintaining index is easier with numeric key.
- Maintain all slowly changing dimenion.
Data warehouse Design
The data warehouse design essentially consists of four steps, which are as follows:
- Identifying facts and dimensions
- Designing fact tables
- Designing dimension tables
- Designing database schemas
Types of database schemas
There are three main types of database schemas:
1. Star Schema, 2. Snowflake Schema and
3. Starflake schema.
Star Schema
- It is the simplest form of data warehouse schema that contains one or more dimensions and fact tables
- It is called a star schema because the entity-relationship diagram between dimensions and fact tables resembles a star where one fact table is connected to multiple dimensions
- The center of the star schema consists of a large fact table and it points towards the dimension tables
- Fact Table = Highly Normalized Dimension Table = Highly denormalized.
- It can be very effective to treat fact data as primarily read-only data, and dimensional data as data that will change over a period of time
Advantages:
- Star schema is easy to define.
- It reduces the number of physical joins.
- Provides very simple metadata.
Drawbacks:
- Summary data in Fact tables (such as Sales amount by region, or district-wise, or yearwise) yields poor performance for summary levels and huge dimension tables.
Steps in designing Star Schema
- Identify a business process for analysis (like sales).
- Identify measures or facts (sales dollar).
- Identify dimensions for facts (product dimension, location dimension, time dimension, organization dimension).
- List the columns that describe each dimension. (Region name, branch name, employee name).
- Determine the lowest level of summary in a fact table (sales dollar).
Fact constellation:
Dimension tables will, in turn, have their own dimension tables. In this case, the Store dimension will contain District ids and Region ids, which will reference district and region dimensions of Store dimension, respectively. This Schema is called Fact Constellation Schema.
Snowflake schema
- A snowflake schema is a term that describes a star schema structure normalized through the use of outrigger tables. i.e dimension table hierarchies are broken into simpler tables
- Represent dimensional hierarchy directly by normalizing the dimension tables ie) all dimensional information is stored in third normal form
- This implies dividing the dimension tables into more tables, thus avoiding nonkey attributes to be dependent on each other.
Advantages:
- Snowflake schema provides best performance when queries involve aggregation.
Disadvantages:
- Maintenance is complicated.
- Increase in the number of tables.
More joins will be needed
Snowflake Schema
Starflake Schema
1. combinations of denormalized Star and normalized Snowflake schemas.
Star Schema vs Snowflake Schema
Slno
|
Star Schema
|
Snow Flake
|
1.
|
Dimension table will not have any parent table
|
Dimension table will have one or more parent tables
|
2.
|
Hierarchies for the dimensions are stored in the dimensional table itself
|
Hierarchies are broken into separate tables in snow flake schema
|
Granularity
- Means what detail data to be stored in fact table
- Types of Granularity
- Transactional Level Granularity
- Periodic Snapshot Granularity
Transactional Level Granularity
- Mostly used
- Each and every transaction stored in fact table
- Drill down and drill up analysis can be done
- Disadvantage 1. Size increases.
Periodic Snapshot Granularity
- Summarizing data over a period is stored in fact table
- Adv : Faster retrieval (less records)
- Disadv : Detail information not available
FAQ
Hierarchy
- Hierarchies are logical structures that use ordered levels as a means of organizing data.
- A hierarchy can be used to define data aggregation.
Example
- country>city>state>zip
- in a time dimension, a hierarchy might be used to aggregate data from the Month level to the Quarter level, from the Quarter level to the Year level.
Level
A position in a hierarchy. For example, a time dimension might have a hierarchy that represents data at the Month, Quarter, and Year levels.
Operational Data Store
- In recent times, OLAP functionality is being built into OLTP systems which is called ODS (operational data store).
- A physical set of tables sitting between the operational systems and the data warehouse or a specially administered hot partition of the data warehouse itself.
- The main reason of ODS is to provide immediate reporting of operational results if neither the operational system nor the regular data warehouse can provide satisfactory accsee.
- Since an ODS is necessarily an extract of the operational data, it also may play the role of source for data warehouse.
Data Staging Area
- A storage area that clean, transform, combine, duplicate and prepare source data for use in the data warehouse.
- The data staging area is everything in between the source system and data presentation server.
- No querying should be done in the data staging area because the data staging area normally is not set up to handle fine-grained security, indexing or aggregation for performance.
Data Warehouse Bus Matrix
- The matrix helps prioritize which dimensions should be tackled first for conformity given their prominent roles.
- The matrix allows us to communicate effectively within and across data mart teams.
- The columns of the matrix represent the common dimensions.
- The rows identify the organizations business processes.
Degenerated Dimension
Operational control numbers such as invoice numbers, order numbers and bill of lading numbers looks like dimension key in a fact table but do not join to any actual dimension table. They give rise to empty dimension hence we refer them as Degenerated Dimension(DD).
I got a good answer from the above description,but it still requires some more update to be made. Please share more content on MSBI Online Training
ReplyDelete