Data Categorization in Data Science

Data Sets

• Different forms of datasets
• Data within Data Science

Data Categorization

• NOIR Topology: Categorization of data based on properties. NOIR represents: Nominal, Ordinal, Interval, Ratio
• Nominal Scale: Categories without any sort of order or rank. Examples:
  • Gender: (M, F)
  • Blood Groups: (A, B, AB, O)
  • Country Code: (048, 040)
• Binary Scale: A type of Nominal scale with exactly two mutually exclusive categories.
  • Examples:
    • Switch: (On, Off)
    • Gender: (Male, Female)

### Ordinal Scale

• Categories with a natural order or rank. Examples:
  • Education Level: (High School, Bachelor’s, Master's, PhD)
  • Satisfaction Rating: (Very Dissatisfied, Dissatisfied, Neutral, Satisfied, Very Satisfied)
  • Movie Ratings: (1 Star, 2 Stars, 3 Stars, etc.)

Interval and Ratio Scales

• Interval Scale: Data with consistent intervals between values. Examples:
  • Temperature (Celsius or Fahrenheit):
    • 10 degrees Celsius - 0 degrees Celsius = 10 degrees Celsius - 20 degrees Celsius (same interval)
  • Time (measured in hours):
    • 10 am - 9 am = 11 am - 10 am (same interval)
• Ratio Scale: Data with a true zero point. Examples:
  • Height: (0 cm to 200 cm)
  • Weight: (0 kg to 100 kg)
  • Age: (0 years to 100 years)

Types of Datasets

• Record Data:
  • Relational Records: Databases with highly structured data stored in relational tables
  • Data Matrix: Numerical or categorical data displayed in rows and columns (e.g., Spreadsheets)
  • Transaction Data: Transactions from a point of sale system, including timestamps and items
  • Document Data: Text documents represented as Term-Frequency Vectors (a matrix showing word frequency in a document)
• Graphs and Networks: Data represented as nodes (points) and edges (connections) used in network analysis.

Data in Data Science

• Entities: Objects or things.
• Attributes: Measurable properties of an entity.
• Data: Measurement of an attribute. Computers can work with different types of data, such as audio, video, text, etc.

CRISP-DM Methodology

• Stands for Cross Industry Standard Process for Data Mining
• A cycle that describes commonly used approaches for data mining experts

Business Understanding

• Understanding the project objectives and requirements from a business perspective
• Converting this knowledge into a data mining problem definition

Data Understanding

• Initial data collection
• Activities to get familiar with the data
• Identifying data quality problems
• Discovering insights into the data

Data Preparation

• Constructing the final dataset
• Transforming and cleaning data for modeling tools
• Tasks include table, record, and attribute selection

Modeling

• Selecting and applying various modeling techniques
• Calibrating model parameters to optimal values
• Often requires returning to the data preparation phase

Evaluation

• Thoroughly evaluating models before deployment
• Reviewing the steps executed to construct the model
• Ensuring it achieves the business objectives

Deployment

• Creating the model is generally not the end
• Organizing and presenting knowledge in a way that is useful to the customer

Data Collection

• Systematic process of gathering observations or measurements
• Gaining first-hand knowledge and insights

Why Data Collection?

• Analyzing a problem and learning about its outcome and future trends
• Arriving at a solution for a question about the future

Data Collection Methods

• Primary: Original data collected directly from the source
• Secondary: Data collected by another person that is readily available

Primary Data Collection Methods

• Qualitative: Analyzing the quality and understanding reasons behind something
• Quantitative: Expressing data in figures or numbers using traditional or online methods

Secondary Data Collection Methods

• Data from sensors, magazines, and documents

What Before Data Collection?

• Aim of the problem
• Type of data to be collected
• Methods and procedures to collect, store, and process data

Step 1: Define the Aim of the Problem

• Identifying the exact problem you want to solve
• Clearly defining the objectives for data collection

This quiz explores the various forms of data, particularly focusing on the NOIR topology used to categorize data based on properties. It covers different types of scales including nominal, ordinal, interval, and ratio, along with examples for each category. Test your understanding of how data is structured in the context of data science!