Business Intelligence Lecture 2 Summary

Business Intelligence - Lecture 2 Summary

• Model Definition: Models represent portions of a business process, enabling precise formulation of analytical questions.

• Representation Function: Crucial for realizing the model's representation.

• Model Language: Enables the proper formulation of representations.

• Models of Phenomena: These models portray aspects of business processes relevant for analytical purposes.

• Phenomena: Features within the business process of interest for analysis.

• Caricatures: Models simplify phenomena for practical use.

• Idealized Models: Models based on simplified processes like business operations, treatments, or course designs.

• Analogical Models: Employ ideas from other scientific domains e.g., gravity model.

• Phenomenological Models: Statistical models, like regression, are used.

• Models of Data (Machine Learning/Data Mining): Learning the most appropriate models are critical for data analysis.

• Models of Theories: Models representing formal systems (ontologies)

• Understanding of formal systems: Defining theories from data instances.

• Languages for Models: Specific languages like BPMN for process flow modeling exist.

• Formulation of Models: Each language has specific semantic for defining model elements in a specific problem.

• Model Structures (Syntax/Semantic):

  • Syntax: Defines basic elements and their combination rules.
  • Semantic: Defines the meaning of model elements independent of the domain.
  • Notation: Provides a clear way to communicate model expressions.
  • Model Elements: Specific components to represent the business process.
  • Generic Questions: Analytical inquiries regarding model elements.

• Modeling (Conceptual Modeling): Mapping domain semantics into suitable model configurations.

• Model Configuration: Allow for the formulation of analytical questions (admissible expressions).

• Connections with Observations: Model configurations should align with business data observations.

• Model Variability: Addresses the nuances and variations of real-world data. Statistical variability can affect data precision.

• Model Quality (Assessment):

  • Correctness: Model structure and domain interpretation are correct.
  • Relevance: Model function meets intended purposes.
  • Economic Efficiency: Balance between model's complexity and cost.
  • Clarity: Model's ease of understanding.
  • Comparability: Model's ability to function within the overall analysis framework.
  • Objectivity: Results aren't affected by the user's biases.
  • Reliability: Consistent results.
  • Validity: Practical usefulness.
  • Content Validity: Model accurately reflects the essential features.
  • Criterion Validity: Correlation with other known methods.
  • Construct Validity: Ability to derive new insights from the model.

• Models and Patterns:

  • Patterns: Describe local behaviors.
  • Models: Describe broader behaviors.

• Graph Structures (Model Language):

  • Syntax: Includes nodes, directed and undirected edges (connections).
  • Notation: Numerical (e.g., adjacency matrices) and visual representations are employed.
  • Model Elements: Various graph structures (trees, networks, bipartite graphs).
  • Generic Questions: Queries about graph properties.

• Data Generation:

  • Data Types: Transactional, administrative and web data are pertinent.
  • Population Definition: Defining the group or entity represented by the data (e.g. customers, reviews etc.).
  • Measurements: Precisely quantifying data qualities.