Data Science Fundamentals

Attendance

• Mark attendance within a minute.

Data in Data Science

• Entity: A particular thing.
• Attribute: Measurable/observable property of an entity.
• Data: A measurement of an attribute.
• Computers can manage various data types (audio, video, text, etc.).

Data Categorization

• NOIR: Nominal, Ordinal, Interval, Ratio
• Classification scheme for data types.

Types of Datasets (1): Record Data

• Relational records: Database tables with highly structured data.
  • Example tables: "Person" (Pers_ID, Surname, First_Name, City) and "Car" (Car_ID, Model, Year, Value, Pers_ID).
• Data matrix: Numerical matrix, crosstabs. Example includes a table of sales data organized by region/product.
• Transaction data: Example includes a table detailing orders with items and unique order IDs.
• Document data: Term-frequency vector/matrix describing text documents. Example includes a table listing documents with counts/frequencies of words like "team" or "coach".

Types of Datasets (2): Graphs and Networks

• Transportation networks: Maps of transportation routes.
• World Wide Web: Network of interconnected webpages.
• Molecular structures: Network of atoms.
• Social or information networks: Relationships between entities.

Nominal Scale

• Definition: A variable with values from a set of mutually exclusive codes with no logical order.
• Example Codes: Gender (M, F); Blood groups (A, B, AB, O); Country code (048, 040).
• Note: The variable can be numbers, letters, strings in label format. The number of categories should be finite and mutually exclusive.
• Note: Numerical values don't have mathematical meaning.
• Important: Values can't be ordered. (A != B, but A=A).

Binary Scale

• Definition: A special case of nominal scale with two mutually exclusive categories (e.g., ON/OFF, True/False).
• Example: Switch status, attendance (Yes/No).
• Note: A binary variable is only two values.

Symmetric and Asymmetric Binary Scales

• Symmetric: Binary choices have equal importance, e.g., Gender.
• Asymmetric: Binary choices have unequal importance, e.g., medical test outcome (covid positive/negative).

Ordinal Scale

• Definition: Ordered nominal data; the variable generates ordered data.
• Example: Shirt size (S, M, L, XL, XXL).
• Note: Values can be ordered; use relational operators ( <, >, =).

Interval Scale

• Definition: Data measured on a numerical scale with equal intervals between adjacent values.
• Example: Temperature (Celsius, Fahrenheit), IQ scores.
• Note: An interval scale does not have a true zero.

Ratio Scale

• Definition: Data measured on a numerical scale with equal intervals and a true zero.
• Example: Weight (kg), Income (USD),
• Note: All arithmetic operations (addition, subtraction, multiplication, division) are permissible.

Operations on Data

• Distinctiveness: Data values are distinct.
• Order: Data values are ordered.
• Addition: Data values can be added.
• Multiplication: Data values can be multiplied.

Continuous and Discrete Data

• Discrete: Data can only take on certain individual values, e.g., the number of pages in a book.
• Continuous: Data can take on any value within a certain range, e.g., the length of a film.