Info System Exam 2

ERP

Answers What, When, Where, and How; focuses on products, services, and orders

CRM

 Answers Who, What, Where, and When; focuses on customers, products, and contacts

BI (Business Intelligence)

Answers Why, What if, What matters, and Prediction; focuses on products, people, and events.

Core ERP Components

Internal backend modules connected directly to the ERP Database: Accounting, Finance, Marketing, Production, Inventory, and HR

Extended ERP Components

External and customer-facing components integrated with the central ERP system: Business Intelligence, Customer Relationship Management (CRM), Supply Chain Management (SCM), and E-Commerce

Software Architecture

Strategic design and blueprint of the "what" and "why" of software

Software Infrastructure

Strategic design and blueprint of the "how" and "where" of software

Hardware Architecture

Strategic design and blueprint of the "what" and "why" of hardware

Hardware Infrastructure

Strategic design and blueprint of the "how" and "where" of hardware

BI-Survey Finding on Decision-Making

Found that 40% of decisions in organizations were not based on information, and that best-class organizations are more likely to make data-driven decisions

Performance Impact of Lack of Data

Reinforced the notion that lacking data for decision-making directly contributes to poor company performance

Impacts of Gut-Feel Decisions

Leads to wrong decisions, decisions made too late, and decisions made reactively rather than proactively before bad events happen

Technology/Software Diversity

A reason for gut-feel decisions where information is stored across different software systems, making it inaccessible

Outdated Technology

 A cause of gut-feel decisions where computers are legacy systems using proprietary hardware

Unintegrated Technology

A cause of gut-feel decisions where disparate systems are not connected or integrated

Unplanned Acquisitions

 A driver of gut-feel decisions where mergers and acquisitions add unnecessary complexity to hardware and software systems

Data Rich But Information Poor

An information hierarchy state typical of gut-feel firms with a massive base of data but only a tiny tip of usable information

Data -

Unstructured symbols and characters at the base of the information hierarchy.

Information

- Structured data (data with who, what, and when) used directly to make decisions.

Knowledge

- Highly structured information located above information in the hierarchy.

Wisdom

- The top level of the hierarchy answering "Why act? What makes sense?".

Structured Decisions

- Frequent, routine decisions with an agreed-upon procedure for resolution where the actor only needs to recognize the issue; made by lower management/operational staff using data.

Semi-Structured Decisions

- Decisions that are partly routine and partly unique, requiring judgment, evaluation, and insight for parts of the problem; made by middle management using information and knowledge.

Unstructured Decisions

- Non-routine decisions without an agreed-upon resolution procedure, requiring full judgment, evaluation, and insight; made by top management using knowledge and wisdom.

Production Environment

- Operational IT environment housing modules like Sales, Marketing, and Production backed by a DBMS and Central ERP Database

Analyses Environment

- IT environment housing a Data Warehouse Database, data marts, a DBMS, and Business Intelligence applications (BI software and Reportings vs. software).

ETL Software (Extract, Transfer/Transform, Load)

- Software used to move data from the production environment (ERP/databases) into the analytical environment (Data Warehouse).

Data Warehouses

- Special databases used to manage, prepare, consolidate, and store data extracted from other sources, intended specifically for data mining and data analyses.

Data Mart

- A subset of an organizational data store usually oriented to a specific purpose, designed to focus on specific business functions or communities, and often derived from a data warehouse (e.g., Web sales, Customer, Inventory data marts).

Operational Databases (Purpose)

- Created to store large volumes of operational transaction data.

Data Warehouses (Purpose)

- Designed and developed to enable data analytics and deep data mining.

Operational Databases (Level of Detail)

- Stores very detailed information about hourly events with high granularity.

Data Warehouses (Level of Detail)

- Stores aggregated information about hourly, weekly, monthly, etc., events with low granularity.

Operational Databases (Timeliness)

- Changes are updated in real-time, making the information timely.

Data Warehouses (Timeliness)

Changes are updated periodically (e.g., every other day, weekly), making it not too timely.

Operational Databases (Span of Time)

- Covers a period of 3 years or less

Data Warehouses (Span of Time)

- Typically spans a period of over 3 years or more.

Operational Databases (Size)

- Smaller in size compared to data warehouses.

Data Warehouses (Size)

- Typically larger in size than operational databases.

Types of BI Software

- Categories supporting decision-making: Reporting Systems and OLAP Systems (used to Assess), and Data Mining Systems / predictive analytics (used to Predict).

Simple Reporting Systems

- Tools that generate information from data through four operations: filtering data, sorting data, grouping data, and making simple calculations on the data.

Primary Use of Reporting Systems

- Used primarily for assessment to address questions like: What has happened in the past? What is the current situation? How does the current situation compare to the past?.

Reporting Systems Workflow

- Transforms raw transactional data (e.g., NASDAQ trades) via a reporting tool into structured reports/information and visual dashboards.

Benefits of Reporting Systems

- Users get quick summaries of data, quick charts, and the applications are easy to learn.

Disadvantages of Reporting Systems

- Users need many reports to answer questions, and it is not possible to define relationships between data (data models).

OLAP (Online Analytical Processing)

- Complex reporting software that builds on the functions of a reporting system by providing an OLAP cube and modeling to determine data exploration and analysis.

Dynamic OLAP Reports

- Interactive reports that allow users to drill down the data hierarchy and explore different views of data within the OLAP cube.

"Online" in OLAP

- Refers to the capability of the system to provide real-time manipulation of data.

OLAP Cube (Dimensions)

 - A multidimensional data model (e.g., Region, Products, Time) used to analyze business metrics across different perspectives.

OLAP Cube: Product Analysis

- Slicing and viewing data across dimensions to analyze how products performed across regions (e.g., discovering a product did not do well in the North West region).

OLAP Cube: Regional & Time Analysis

- Slicing and viewing data across dimensions to analyze how regions performed across time (e.g., discovering the North West region had a slump in 2019).

OLAP Cube

- Presentation of dimensions with associated measures.

OLAP Report (Capabilities)

Reports created and published by data analysts to answer global questions, or created by users pulling data themselves to answer their own questions, alter formats, drill down into data, and divide data into more detail

Dimension

The business category (business entity) of a measure; examples include product, customer, customer location, and sales region

Measure

- The attribute of a dimension (the data item of interest) that is summed, averaged, or otherwise classified by specific categories/dimensions; for example, dollar ($) value of sales.

OLAP Systems Architecture (Components & Flow)

- Takes raw data sources through a data model into an OLAP cube, feeding outputs such as self-service reporting & analysis, dashboards/scorecards, pixel-perfect reports, standard reports, and threshold-based alerts.

Provisioning OLAP BI

- Delivery methods for OLAP BI, including local desktop, local area network/wide area network (LAN/WAN) server, mobile server, Internet server, and Cloud server.

OLAP Application: Global Bookstore Chain

- Designing an OLAP system to control chain results by selecting relevant dimensions (e.g., location, book genre), dimension levels (e.g., country > city > store), and measures (e.g., total sales, units sold).

ERP, E-Commerce & BI Integration

- Unified enterprise architecture connecting ERP applications/database, Ecommerce applications/database, and Web servers with a central Data Warehouse, BI software server, and Internet Gateway Server across internal networks and external clouds.

Stage 1: Leveraging Data

- Setting up architecture and infrastructure to capture, manage, and query large volumes of structured operational data; stores structured data using SQL databases and tools.

Stage 2: Leveraging Data

- Setting up architecture and infrastructure to capture, manage, and analyze medium volumes of internally selected and structured/semi-structured data; stores structured and semi-structured data with centralized data warehouses, OLAP, and data mining tools.

Stage 3: Leveraging Data

- Setting up architecture and infrastructure to capture, manage, and analyze large volumes of internally and externally selected, aggregated raw, cleaned, unstructured, and semi-structured data; uses distributed storage/processing data warehouses and sophisticated analytical tools.

Reporting Systems (Analytics Type)

- Descriptive analytics used for supporting decision-making.

OLAP Systems (Analytics Type)

- Diagnostic analytics used for supporting decision-making.

Data Mining Systems (Analytics Type)

- Predictive analytics (also called predictive analytics) used for supporting decision-making.

Data Mining (Definition & Alternate Name)

- Machine learning/data mining software technologies that enable firms to uncover meaningful, but hidden patterns and relationships within data; also known as Knowledge Discovery in Data (KDD).

Data Mining (Prediction Example)

- Estimating the probability that a customer will default on a loan.

Analytic Tools vs. Reporting Tools

- Analytic tools process data using sophisticated and mathematically complex techniques to find patterns and make predictions, contrasted with simple operations like sorting, grouping, and summing used by reporting tools.

Data Warehouses (Additional Contents)

- Special databases intended for data mining and data analyses that also include consumer data purchased from other vendors and meta data.

Unsupervised Data Mining

- Application of analyses to data and observing the results thereafter without prior guidance; examples include clustering, decision analyses, and market basket analyses.

Supervised Data Mining

- Application of analyses in which the data analyst uses their knowledge to influence what model is used for the analyses; examples include regression analyses and neural network (neuronal network) analyses.

BI Prediction Process (Step 1a & 1b)

- Splitting sample customer data into training sets (customers with & without expected behavioral outcome) and test sets (other customers with unknown behavioral outcomes).

BI Prediction Process (Step 2)

- Examining data with two or more modeling/analytic techniques to evaluate model performance (e.g., comparing Model A, Model Q, and Model K).

BI Prediction Process (Step 3)

- Training and examining data with the best-performing model to extract insights and behavior-influencing predictors (e.g., Tour De France hours, Motocross fan, gender, preferred movie).

BI Prediction Process (Step 4)

- Examining data with the best predictive model and predictors to generate results (e.g., predicting specific customers likely to buy).

Big Data (Growth Context)

- 90% of organizational data was created in the last two years, and the amount of data available is doubling annually.

Big Data (Definition)

The collection, storage, and analysis of extremely large, complex, and often unstructured data sets used by organizations to generate insights that would otherwise be impossible.

The Three Vs of Big Data

- The three distinguishing characteristics from conventional data: Volume (huge volume), Velocity (grows or changes rapidly), and Variety (contains different varieties of data types).

Small Data

- Data from enterprise applications and operations stored in a fixed format in transactional databases.

Large Data

- Aggregated data from enterprise applications and similar sources stored in star format in analytical data databases (e.g., data warehouses).

Big Data (Storage Format)

- Disaggregated data from multiple sources and applications stored raw and in their original formats in storage and cataloging data warehouses (e.g., data lakes).

Data Lake

- A catch-all term for storage and access technologies used in Big Data that allow storage of data in both structured as well as raw, unfiltered formats, providing tools to "pipe out", filter, and refine data into information.

Applications Leveraging Big Data

- Big Data Analytics use cases (applying OLAP and Data Mining to identify hidden patterns), Machine Learning use cases, and Artificial Intelligence use cases.

Artificial Intelligence (AI)

Computer software that can mimic or improve upon functions that would otherwise require human intelligence; also serves as the historical root of data mining.

Enablers of AI Spread

- An explosion of tools fueling AI adoption: new generation of hardware chips, cloud resources, open source algorithms, software development kits (SDKs), and data-capture tools.

Machine Learning: Recognizing Shapes

- Models trained to recognize a shape and classify unknown objects as conforming to the shape or not; used in brain and medical imaging applications.

Machine Learning: Generating Shapes

- Generative learning models trained on shapes/images to produce unique images or text descriptions (e.g., a cat) on demand from a user's text prompt.

Machine Learning: Generating Topic-Related Text

- Models trained on billions of types of text to produce unique configurations of outputs on demand when given a text prompt.

Machine Learning: Generating Music

- Models trained on billions of scores of music to produce unique musical compositions on demand when given a text prompt.

Supervised Learning

- Machine learning models that learn from labeled data (known outcomes) and improve by comparing predictions with actual results.

Unsupervised Learning

- Machine learning models that learn by identifying patterns in unlabeled data without any guidance.

Reinforcement Learning

- Machine learning models that learn by interacting with their environment and receiving rewards or penalties.

Semi-Supervised and Self-Supervised Learning

- Machine learning models that learn by combining both labeled and unlabeled data to learn and find patterns.

Data Mining vs. Machine Learning (Data Mining Focus) -

Process of extracting useful information from vast amounts of data to discover new, accurate patterns; works on raw data, relies on human intervention, and is ultimately created for use by people.

Data Mining vs. Machine Learning (Machine Learning Focus) -

A form of artificial intelligence where algorithms learn from data, with or without explicit human guidance, to improve predictions or classifications; works with algorithms and can "teach" itself without depending on human influence.

AI Autonomy

- Can be autonomous or semi-autonomous (particularly in advanced systems); traditional AI models generally lack the ability to learn or adapt after initial programming and rely on predefined rules/algorithms (e.g., traditional chess-playing AI).

Machine Learning Autonomy

- Inherently autonomous systems that constantly evolve, adapt, and refine their models based on processed data without human input, making them more flexible than traditional AI.

AI Decision-Making Approach

- Utilizes a broad range of rule-based systems or logic-based reasoning tools to make decisions based on structured facts and relationships.

Machine Learning Decision-Making Approach

- Uses algorithms focused on learning from patterns in data to identify relationships and adjust approaches based on insights

ChatGPT (AI vs. Machine Learning)

- A combination of both AI and machine learning that uses advanced machine learning models to generate human-like responses, all powered by AI.