Business Intelligence Chapter 2

Study Notes

Chapter 2 focuses on descriptive analytics, covering data nature, statistical modeling, and visualization.
Data is a collection of facts, often obtained from experiences, observations, or experiments.
Data types include numbers, words, images, and more.
Data is the foundational element for deriving information and knowledge.
Data quality and integrity are crucial for analytics. Data integrity encompasses accuracy, completeness, consistency, and validity.
Metrics for analytics-ready data include source reliability, content accuracy, accessibility, security/privacy, richness, consistency, currency/timeliness, validity, and granularity.
Data is categorized into structured, unstructured, and semi-structured forms.
Structured data is standardized, follows a format, and is easily accessed (e.g., names, dates, addresses, etc.).
Unstructured data includes any combination of text, images, voice, and web content.
Semi-structured data falls between structured and unstructured, with some organizational structure (e.g., XML, JSON, log files).

Categorical variables represent types or groups (e.g., race, sex, age group).
- Nominal data are used for labeling without quantitative value (e.g., gender, color).
- Ordinal data have an inherent order (e.g., Likert scale, educational level).
Numerical variables represent measured values that can be logically ordered.
- Interval data have order and difference between values (e.g., temperature).
- Ratio data have order, difference, and a meaningful zero point (e.g., height, income).

Real-world data is often dirty, requiring preprocessing for analytics.
Data preprocessing involves data consolidation, cleaning, transforming, and reduction.
- Data reduction techniques include dimensional reduction and variable selection for variables and sampling/stratification for cases/samples.

Statistics is a set of mathematical techniques to characterize and interpret data.
- Descriptive statistics describe data, while inferential statistics draw inferences about a population from sample data.
Measures of central tendency include the arithmetic mean, median, and mode.

Dispersion, or variability, measures the spread or variation in a given variable.
Measures of dispersion include range (max - min), variance, standard deviation, and mean absolute deviation (MAD).

Data visualization uses visual representations to explore, make sense of, and communicate data.
Visualizations can span from histograms to graphs, charts, illustrations, etc.
Visual analytics combines information visualization with predictive analytics techniques.
Performance dashboards are used for combining and visualising key information from multiple sources.

Regression is a technique in statistics used to understand or model the relationship between variables.
Regression can be used to build models that allow for prediction and analysis of data.
It typically involves determining the explanatory (input) and response (output) variables.

Reports translate information into actionable decisions.
Reports involve various functions, such as communication, maintaining departmental efficiency, providing analysis results, persuasion, and knowledge management for the organization.
Business reports may vary in format, distribution, and source of information.
Reports can utilize key performance indicators (KPIs), and can use various types of visualizations to support the presentation.