Business IT and Business Intelligence

Introduction to Business IT

• Early commercial computers were large, consolidated systems called "Mainframes."
• Manufacturers produced limited quantities of each model, targeting specific markets.
• IBM's System/360 was an early example of a more flexible and scalable range of compatible computer systems for businesses.
• Mainframes were often replaced by "Minicomputers" in the 1960s and 70s.
• Minicomputers were disrupted by Personal Computers (PCs) in the 1980s.
• Technological advancements lead to increased data volumes.

Business Intelligence (BI)

• BI involves information processing for business performance.
• BI filters information overload and extracts relevant data.
• Management Information Systems (MIS) are used to provide flexible and fast evaluations.
• Data warehouses store information and knowledge.
• In a broad sense, BI includes all applications that directly or indirectly support decision-making processes (storage, processing, evaluation, presentation).
• Narrowly, BI is the core application directly supporting decision-making, especially decision-finding (OLAP, MIS).
• Analysis-oriented BI includes all applications used by decision-makers in direct contact.
• BI is a specific class of IT systems, a collection of technologies and concepts for decision support, a transformation process of data into decision-relevant information, and a collection of IT solutions to support decision-makers with interactive functions.
• An integrated entrepreneurial approach supports IT-based decision-making.

Role of BI in Companies

• At the Top Management Level:
• The highest leadership levels make decisions with uncertainty and existential significance.
• Implementation occurs through subordinate management levels.
• Responsibilities include defining strategic company goals (enforcement, control, coordination of overall company planning).
• At the Middle Management Level:
• Serves as the interface between company leadership and executive units.
• Responsibilities include implementing and fulfilling decisions and goals set by top management (concrete plans and programs).
• Powers include authorization and procuration.
• At the Lower Management Level:
• Serves as the interface to the operative units of the execution system (implements decisions of superiors).
• Responsibilities include planning, management, and control of manageable executing organizational units (quality and efficiency of work processes).
• Coordination and support are necessary in decision-making.
• This involves the coordination of planning, control, and information provision, which is necessary due to the complexity of the management system.

Types of Decisions

• Unstructured decisions: Decision-makers must provide evaluation, assessment, and insights for problem-solving.
• Example: Deciding whether or not to enter a new market.
• Typically made at the Top Management level.
• Semi-structured decisions: Decision-makers solve problems partly answerable through an accepted process.
• Example: Why is the report on order fulfillment in Würzburg declining?
• Typically made at the Middle Management level.
• Structured decisions: Decision-makers solve repetitive and routine problems with a clear solution procedure.
• Example: Should the company cooperate with a specific supplier?
• Typically made at the Lower Management level.

Practical Application and Outlook on BI Software

• Critical capabilities should be considered before investing in an Analytics and BI platform:
• Security, Governance, Cloud-enabled analytics, Data source connectivity, Data preparation, Catalog, Automated insights, Data visualization, Natural language query, Data story-telling, Natural language generation, Reporting.
• Examples of Industry Use: Verizon and MyToys
• Current Trends include:
• User empowerment to expand data management and analytics.
• Modernizing data architecture, organizing data and analytics, utilizing data in processes (e.g., DevOps).

Data Models: Databases

• Data organization used to be done with files.
• Information systems in companies are typically not planned from the outset.
• Each functional area often developed its own systems, encompassing the actual application and associated data organization.

Problems of Data Organization with Files

• Data redundancy and data inconsistency: Data was stored redundantly in each system, leading to errors and inconsistencies.
• Dependency between program and data: Modification of data structure necessitated program changes, lacking encapsulation.
• Lack of data exchange and flexibility: Insufficient integration made the realization of organization-wide processes difficult.
• Lack of data security

Database vs. Database Management System

• A database is a collection of data describing a segment of the real world. Changes in the real world segment are reflected in the database.
• A database is a logically coherent collection of data with a specific inherent meaning (not a random data collection).
• It designed, developed, and filled for a specific purpose.
• A Database Management System (DBMS) is a software system allowing users to create and maintain a database.

Characteristics of the Database Approach

• Data isolation and abstraction:
• Internal data management is fully hidden by the DBMS, so structural changes do not require alterations in accessing applications.
• Database catalog ("Data Dictionary"):
• It contains a directory of all data structures, types, and constraints regarding permissible value ranges, not just the database itself.
• Support for multiple data views:
• The creation of "Views" allows different perspectives of the database for different user groups (e.g., through aggregation or filtering of data).
• Multi-user access and -control with user roles:
• Database administrators assign access rights, coordinate and monitor usage, and procure SW/HW resources.
• Database designers determine data to be stored in the DB, its structures, and necessary views and their notation.
• System Analysts & Application Programmers capture user requirements, specify transactions on the data, and implement them as software.
• End users also benefit.

Advantages of a DBMS

• Redundancy control:
• Data is stored internally only once and varying requirements for data and its preparation are achieved using Views.
• Access restrictions:
• A DBMS has a security and authorization subsystem through which the administrator can define access rights and restrictions.
• Persistent Storage:
• DBMS enable the permanent storage of data and derivatives and actions can derive other data from data through rules.
• Multi-user interfaces:
• Interfaces come in graphical or textual form, as well as programming interfaces for access from applications.
• Relationships between data:
• DBMS consider relationships between data types.
• Integrity conditions:
• DBMS allow the definition of semantic conditions on the data structure.
• Restoration of the DB:
• A DBMS can reset the DB to a correct state.

Disadvantages of DBMS

• Additional costs of DBMS usage are due to:
• high initial investments for HW, SW, and training.
• the universality of a DBMS for defining and processing data.
• additional costs for security, concurrency control, recovery, and integrity.
• Using conventional files may be preferable if:
• Data and applications are simple, clearly defined, and unlikely to change.
• Some programs have real-time requirements not met by a DBMS.
• Multi-user access is not required.

Classification of DBMS

• Numerous DBMS products are available that differ in:
• Data model (e.g., relational).
• Number of users (single vs. multiple).
• Database distribution (central vs. decentralized).
• Storage of files.
• Application specificity.
• Performance, and Costs.

ER Data Models

• Data structuring within a database or through a DBMS is based on "data models"
  • Conceptual data models offer concepts for describing the real world
  • A physical data model which describes the details of the data storage
• Between these two model types exist logical data models.
• These models provide concepts that end-users can understand while being closely related to how data is organized on the computer.

Conceptual Data Modeling

• Describes the arrangement of data items in the database for the user.
• The general procedure involves identifying relationships between different data items, identifying redundant data items, and deriving an efficient arrangement or grouping of data items to fulfill operational information needs.
• Elements of ER Models include:
• Entity is a phenomenon about which data is stored and maintained.
• Attributes like name or age
• A Relationship between two or more entities represents a connection between the entities
• ER models are abstractions of reality. Attributes. Simple (atomic) and composite attributes, which can be divided into smaller parts
• Single-valued or multi-valued attributes:
• Most attributes have a single value but can sometimes accept multiple values.

Attribute Considerations

• A range can specify a lower and upper limit for the number of values an attribute can have.
• Stored or derived attributes:
• Attributes can be related, like age and birthdate, where one can be derived from the other. The attribute from which the value is derived is the stored attribute.

Null Values

• Some attribute values may be missing in particular instances and such attributes are assigned as null values

Entity Sets

• The entirety of all entities of an entity type. A database typically contains groups of similar entities that have shared attributes.
• An entity type defines a group of similar entities
• A Key attribute has values are unique for each entity and these have underlined names

Relationship types

• Relationships can exist. These relationships are present in ER Models and can have their own attributes. The degree relates to the number of entity types involved.

Cardinal Relationships

• Cardinal Ratios: Specifies how many relationships an entity can participate in.
• Participation Constraints: Specifies whether an entity must participate in relationship with another entity.

Relationship Type Attributes

• Relationship types can also be characterized by properties through attributes.
• You can model freely if an attribute relates to a relationship or entity.
• attributes can alternatively be assigned to one of the entity types involved in a 1:1 relationship, while for 1:N relationships, must be assigned to the N-fold entity type.
• Weak Entity Types: Entity types that do not have their own key attributes. Must always add a weak type.

Relational Data Modeling

• Logical DB Models as ER Models are strongly user oriented, are very user side and independent in terms of technical implementation in DBMS.
• Problems: There are no DBMS products based on the ER approach, so ER Models must then be transformed into a logical data model

Phases of Data Modeling

• Conceptual design
• Logical design
• Physical Design. Relational Data Model

Concepts of Relational Models

• Database as collection of relations (= tables).
• Each row represents a list of related values (= data record).
• The Row is referred to as tuple
• Column is referred to as attribute.
• Attributes have data type, format, and defined value ranges
• Relation database consist of relations and schemas. Where a relational database can be broken into relation schemas together with integrate conditions

Important considerations regarding this include:

• Key Attributes: identifying attribute which define uniqueness. Usually, a primary key candidate must be determined.
• Foreign keys that reference other key-attributes which are used to create joins for relationships. The integrity of these must be preserved.
• Transformations must reduce redundancy in databases.

Transformation of ER-Models

• Transformation of ER models into relational models is a key process in relational database design
• There is a problem with modelling, the more complete modelling will produce many smaller relations. This means questions take longer to process

Structured Query Lanuage

• DBMS provides a direct form accessing this data, and the interface include functions for organization and database processing

• The languages used are Data Definition Language (DDL) and Data Manipulation (DML).

• Data Query Language (DQL)

• The SQL language is generally used for any application that meets the specification and requirements of the DDL, DML AND DQL

Structured query language SQL

• Support all important relational DBMS.
• Provide user queries or as interface
• The generic structure is SELECT spaltename1, FROM tabelle1,WHERE Bedingung
• Relational algebra is applied

Definitions

• Selection refers to only taking certain value
• Projection sets which attributes are taken upon to be results
• The resulting is then added together

Joining tables SQL

• Equi-join uses a cartesian product with relations to identify the correct tuples and their relations. Where, it requires the content of columns to be identical to the column in question.

Table Relationships in SQL

• The natural join, this cartesian product builds the relations for the table
• After that, only specific content is displayed, such as common or unique values
• Types of joins:
• Left (Outer) Join
• Right (Outer) Join
• Full (Outer) Join

General Concepts

• data Warehouse(DW) assist in supporting organization structured on a particular form. So they are available to query and report.

Definitions of data warehouses

• Subject oriented data is organized by topic
• Is an integrated part of data from system
• Data is related to time
• Data must be non volatile because is already in the data warehouse
• Uses a client server architecture

Different Levels

• Level is the most detailed level the information can be grouped. The data mart is a sub set that has data the is related to department.

The DW Core

• Metatdata where data is stored depending on how it is interpreted

• The Middlewhere layer or Translation Vermittlungsschicht

• The aim of designing a good architeture to ensure the architecture is easily changed especially in relation to scaling.

• Datawarehouse is the basis of a One level archicture architecture or Multi level (Mutli-tier) providing scalability

Data Architectures

• You need to decide on a DBMS and decide what to generate.
• A data warehoue must meet the needs of the data and analysis

Types of Data Modelling

• Paradigms build based on different bill and comparison with architecture

Analysis factors of Datamodels

• Need for architecture
• Need for end user
• compatibility with systems

Multidimensional data structures

• Models are made in free structures or data sets are inserted, deleted and switched out.
• There are parallel dimensions that allow additional layers of modelling which aid with structres.
• Operations are performed across these layers by specific roles with specific permissions often across teams

The development of these occurs across levels that must be determined.

• These depend on business needs and also technical needs to consider.
• Business are in charge of design and models aid with design with aspects
• This process is then specified and granularized
• The factors that limit granuality are often trade offs with the data volume available.

Multi dimentional Data Models

• Must have the same definitions accordin the following
• Step 1: Business processes are identified.
• Step 2: Granularity is determined based on information
• Step 3:Dimensions are set
• Step 4:Key Measures
• The result is more efficient Business Intelligence gathering.
• The ETL provides a more structured environment for data collection in processes.

ETL

• Includes the main processes, as well as data management processes.
• The major design is the extraction or data, including integration, and the implementation.
• In the implementation different ETL toolsets are included with their strengths and weaknesses.
• A major part is defining the data integration which helps to improve performance.

ETL - Data Preperation

• Filtering and cleaning of data errors must be addressed.
• There are also manual processes to assess data.
• The whole process is Harmonized
• Harmonization allows for more unified approaches to the data. The resulting process is now enriched and structured, and the data is integrated.

Explore the evolution of business IT from mainframes to PCs and the role of Business Intelligence (BI) in information processing. Learn about Management Information Systems (MIS) and data warehouses for effective decision-making. Discover how BI supports various aspects of business decision processes through storage, processing, evaluation and presentation.