Business Intelligence Chapter 2 Summary

Study Notes

Data: A collection of facts, usually obtained from experiences, observations, or experiments. Data can be numbers, words, images, etc. It's the fundamental building block for information and knowledge. Data quality and integrity are critical for analysis. Data integrity includes accuracy, completeness, consistency, and validity of an organization's data.

Data types include structured, unstructured, and semi-structured.
- Structured data: Standardized format, well-defined structure, follows a consistent order, and easily accessed. Examples include names, dates, addresses, credit card numbers, stock information, geolocation. Usually easily understood by machine language.
- Unstructured data: Any combination of textual, imagery, voice, and web content—not organized in a predefined format.
- Semi-structured data: Data that doesn't conform to a rigid schema but has certain organizational elements (like tags or labels). Examples include XML, JSON, HTML, log files.
Data is categorized by properties:
- Categorical data: Represents different types or groups.
  - Nominal: Labels variables without quantitative value. (e.g., male/female, hair color, nationality).
  - Ordinal: Variables have an order, but the differences between the values are not meaningful. Examples include Likert scales.
- Numerical data: Represents values that can be measured and ordered.
  - Interval: Concerned with both the order and difference between variables. (e.g., classifying people as teenager, youth, etc.)
  - Ratio: Deals with the order of variables, differences between them, and has a meaningful zero point. (e.g., income, height, weight)

Data source reliability: Is the source dependable?
Data content accuracy: Does the data match the task/job?
Data accessibility: Is the data easy to access when needed?
Data security and data privacy: Who has access to the data and under what conditions?
Data richness : Is the data complete or near complete?
Data consistency: Is the data collected and combined accurately?
Data currency/timeliness: Is the data up-to-date, recent, and new?
Data validity: Does the data match the expected data?
Data relevancy: Are the variables in the data relevant to the study?
Data granularity: Are the variables and data defined at the lowest level of detail possible?

The real-world data is often not ready for analysis due to its dirty, misaligned, complex, inaccurate nature.
Data preprocessing techniques are needed to ready the data for analysis— including data consolidation, cleaning, transformation, and reduction.
Data reduction: Focuses on reducing the dimensionality (number of features) or the volume of data.
- Variable selection: Selecting relevant variables from a large set.
- Dimensional reduction: Reducing the number of variables.
- Sampling: Selecting a representative subset of data.
- Balancing/stratification: Reducing imbalances in data sets. (e.g., stratified sampling)
Discretization: Converting continuous data into discrete intervals or categories (e.g., age groups, income brackets).
Data normalization: Reorganizing data to remove unstructured and redundant information, standardizing the data format.
Creating attributes: Define attributes to store more information (like priority levels).

Descriptive Analytics: Describes the data as it is (e.g., using mean, median, mode, histograms , skew, kurtosis)
Predictive Analytics: Focuses on understanding trends and patterns to predict future outcomes.
Prescriptive Analytics: Recommends actions based on predictions.
Statistics: A collection of mathematical techniques to characterize and interpret data
Descriptive statistics: Describing the data. (mean, median, mode, measures of dispersion, histograms, skew, kurtosis)
Arithmetic Mean
Median
Mode
Range
Variance
Standard Deviation
Mean Absolute Deviation (MAD)
Inferential statistics: Drawing inferences about the population based on sample data (e.g., regression modeling).

Regression: A technique in inferential statistics used to identify relationships between explanatory (input) and response (output) variables. Can used for hypothesis testing and forecasting.

Report: Information leading to decision-making, a communication artifact conveying specific information. Reports fulfill multiple functions, ensuring proper departmental functioning, providing information, providing analysis results, persuading action, and creating organizational memory.

Metric management reports: Use KPIs and metrics to manage business performance.
Dashboard-type reports: Graphical displays of performance indicators.
Balanced scorecard-type reports: Management system that improves business processes by analyzing internal and external outcomes.

The use of visual representations to explore, make sense of, and communicate data. Involves aggregation, summarization and contextualization of data. Related to information graphics scientific visualization and statistical graphics.
Using Charts, graphs, illustrations, etc.
Different types of graphs and charts include histograms, pie charts, scatter plots, line graphs.

Common in BPM software and BI platforms
Provide visual displays of important information on a single screen for quick understanding and drilling down for analysis.
Effective dashboards should be designed with clear presentation of data and exceptions enabling quick assimilation.
Include design elements like prioritizing alerts, clear use of visuals, information layers (monitoring, analysis, and management).