DATA SCIENCE FUNDAMENTALS
Study Notes • Definitions • Workflow • Metrics • MCQs
1. What Is Data Science?
Data science is an interdisciplinary field that uses statistics, programming, scientific methods, and subject knowledge to
extract useful information from data. It supports understanding, prediction, and decision-making.
• Statistics helps describe data and reason about uncertainty.
• Computer science provides programming, databases, and computing systems.
• Machine learning finds patterns that can support predictions.
• Domain knowledge helps interpret results in a real-world context.
2. Types of Data
• Structured: organized in a fixed format, such as rows and columns in a spreadsheet.
• Semi-structured: contains labels or markers without a rigid table, such as JSON or XML.
• Unstructured: text, images, audio, and video without a predefined tabular structure.
• Numerical: number-based values; discrete values are countable, while continuous values can take a range of values.
• Categorical: labels or groups; nominal categories have no natural order, while ordinal categories do.
3. The Data Science Workflow
• Define the problem: create a clear question and measurable objective.
• Collect data: gather relevant data from suitable, permitted sources.
• Clean data: address missing values, duplicates, inconsistent formats, and errors.
• Explore data: use summaries and charts to understand distributions and relationships.
• Prepare data: transform variables into useful formats for analysis or modeling.
• Build a model: select a suitable statistical or machine-learning method if needed.
• Evaluate: assess results with appropriate metrics and separate test data.
• Communicate: explain findings, assumptions, limitations, and next steps.
The workflow is iterative: exploration may reveal missing information, and evaluation may show that the model or original
question needs revision.
4. Data Cleaning and Preparation
• Investigate missing values before deciding whether to remove, estimate, or flag them.
• Check duplicates carefully; repeated records may sometimes represent genuine events.
• Standardize units, dates, spelling, and category labels.
• Investigate outliers rather than automatically deleting them.
• Scale numerical features when the chosen algorithm is sensitive to different ranges.
• Encode categorical variables when a model requires numeric input.
Data Science Fundamentals • Study Notes Page 1
Study Notes • Definitions • Workflow • Metrics • MCQs
1. What Is Data Science?
Data science is an interdisciplinary field that uses statistics, programming, scientific methods, and subject knowledge to
extract useful information from data. It supports understanding, prediction, and decision-making.
• Statistics helps describe data and reason about uncertainty.
• Computer science provides programming, databases, and computing systems.
• Machine learning finds patterns that can support predictions.
• Domain knowledge helps interpret results in a real-world context.
2. Types of Data
• Structured: organized in a fixed format, such as rows and columns in a spreadsheet.
• Semi-structured: contains labels or markers without a rigid table, such as JSON or XML.
• Unstructured: text, images, audio, and video without a predefined tabular structure.
• Numerical: number-based values; discrete values are countable, while continuous values can take a range of values.
• Categorical: labels or groups; nominal categories have no natural order, while ordinal categories do.
3. The Data Science Workflow
• Define the problem: create a clear question and measurable objective.
• Collect data: gather relevant data from suitable, permitted sources.
• Clean data: address missing values, duplicates, inconsistent formats, and errors.
• Explore data: use summaries and charts to understand distributions and relationships.
• Prepare data: transform variables into useful formats for analysis or modeling.
• Build a model: select a suitable statistical or machine-learning method if needed.
• Evaluate: assess results with appropriate metrics and separate test data.
• Communicate: explain findings, assumptions, limitations, and next steps.
The workflow is iterative: exploration may reveal missing information, and evaluation may show that the model or original
question needs revision.
4. Data Cleaning and Preparation
• Investigate missing values before deciding whether to remove, estimate, or flag them.
• Check duplicates carefully; repeated records may sometimes represent genuine events.
• Standardize units, dates, spelling, and category labels.
• Investigate outliers rather than automatically deleting them.
• Scale numerical features when the chosen algorithm is sensitive to different ranges.
• Encode categorical variables when a model requires numeric input.
Data Science Fundamentals • Study Notes Page 1
