Glossary¶

This Module 4 glossary defines common terms used in notebooks, exploratory data analysis (EDA), pandas, and related Python tools.

Section 1. Notebook Basics¶

Notebook¶

A notebook is an interactive document that combines code, output, text, tables, charts, and notes in one file. In data work, notebooks are often used for exploration, documentation, and early analysis.

Jupyter Notebook¶

A Jupyter Notebook is a common notebook format used for Python, data analysis, visualization, and scientific computing. Jupyter notebooks usually have the file extension .ipynb.

Markdown Cell¶

A Markdown cell contains formatted text, not Python code. Use Markdown cells to write headings, explanations, instructions, observations, and conclusions.

Code Cell¶

A code cell contains Python code that can be run. When a code cell runs, it may produce printed text,tables, charts, logs, or other output.

Cell Output¶

Cell output is the result shown after running a code cell. Output may include text, tables, error messages, charts, or images.

Run a Cell¶

To run a cell means to execute the code or render the Markdown in that cell. Running a Python cell sends its code to the active Python environment.

Run All¶

Run All means running every cell in the notebook from top to bottom. Notebooks can become inconsistent if cells are run out of order.

Kernel¶

The kernel is the running Python process connected to the notebook. The kernel remembers variables, imports, functions, and objects created while the notebook is running.

Restart Kernel¶

Restarting the kernel clears the notebook's active memory. This is useful when checking whether a notebook runs correctly from a clean start.

Python Environment¶

A Python environment is the set of Python packages and tools available to a project. Our project environment comes from the .venv folder created by uv.

Notebook State¶

Notebook state refers to the current memory of the running notebook kernel. A notebook may appear to work because of old variables still stored in memory. Restarting the kernel and running all cells helps detect this problem.

Reproducible Notebook¶

A reproducible notebook can be run from top to bottom and produce the expected results. A notebook is more reproducible when imports, configuration, data loading, cleaning, analysis, and visualizations appear in order.

Section 2. Exploratory Data Analysis¶

Exploratory Data Analysis¶

Exploratory Data Analysis, or EDA, is the process of examining a dataset to understand its structure, quality, patterns, and limitations. EDA usually happens before formal modeling or final reporting.

Dataset¶

A dataset is a collection of data used for analysis. A dataset may come from a CSV file, database, API, spreadsheet, package, or built-in library source.

Observation¶

An observation is one recorded item in a dataset. In a table, each row usually represents one observation.

Variable¶

A variable is a measured or recorded feature. In a table, each column usually represents one variable.

Row¶

A row is a horizontal record in a table. Each row usually represents one observation.

Column¶

A column is a vertical field in a table. Each column usually represents one variable.

Tabular Data¶

Tabular data is data arranged in rows and columns. Spreadsheets, CSV files, and database tables are common forms of tabular data. Python often uses the pandas (panel data set) package to work with tabular data.

Data Dictionary¶

A data dictionary describes the columns in a dataset. It often includes column names, data types, meanings, valid values, units, and missing-value information.

Data Quality¶

Data quality refers to how suitable the data is for analysis. Common data quality issues include missing values, duplicates, inconsistent categories, invalid values, and incorrect data types.

Missing Value (NaN)¶

A missing value is a value that is absent, unknown, or not recorded. In pandas, missing values often appear as NaN for not a number.

Duplicate Row¶

A duplicate row is a row that appears more than once in a dataset. Duplicates may be valid, accidental, or evidence of a data collection problem.

Outlier¶

An outlier is a value that is unusually far from most other values. Outliers may be errors, rare events, or important signals.

Distribution¶

A distribution describes how values are spread across a variable. For numeric data, a distribution may show center, spread, skew, and unusual values.

Summary Statistics¶

Summary statistics are numeric summaries of a variable. Common examples include count, mean, median, standard deviation, minimum, maximum, and quartiles.

Descriptive Statistics¶

Descriptive statistics is a general term for metrics that summarize what is present in a dataset. They describe the dataset but do not, by themselves, prove causes or make predictions.

Count¶

Count is the number of values or observations. In pandas, count often excludes missing values.

Mean¶

The mean is the arithmetic average. It is calculated by adding values and dividing by the number of values.

Median¶

The median is the middle value after values are sorted. The median is often less affected by outliers than the mean.

Standard Deviation¶

Standard deviation measures how spread out numeric values are around the mean. A larger standard deviation means values vary more.

Minimum¶

The minimum is the smallest value in a variable.

Maximum¶

The maximum is the largest value in a variable.

Range¶

The range is the difference between the maximum and minimum values.

Quartile¶

A quartile divides sorted numeric data into four parts. The 25^th percentile, median, and 75^th percentile are common quartiles.

Correlation¶

Correlation measures the strength and direction of a linear relationship between two numeric variables. Correlation does not prove causation.

Correlation Matrix¶

A correlation matrix is a table showing correlations among multiple numeric variables. The diagonal values are always 1.0 because each variable is perfectly correlated with itself.

Grouping¶

Grouping means splitting data into categories before calculating summaries. For example, penguin measurements might be grouped by species.

Aggregation¶

Aggregation means calculating summary values for groups of data. Common aggregations include count, mean, minimum, maximum, and standard deviation.

Cleaned Data¶

Cleaned data is data that has been adjusted or filtered to support analysis. Cleaning may include removing missing values, correcting types, renaming columns, or selecting relevant rows.

Clean View¶

A clean view is a modified version of the data used for analysis while preserving the original dataset. A clean view helps avoid accidentally changing the raw data.

Section 3. Python Basics for Notebooks¶

Python¶

Python is the programming language used in the notebook. It is widely used for data analysis, automation, visualization, machine learning, and scientific computing.

Import¶

An import brings external code into the current Python session. For example, import pandas as pd makes the pandas library available using the name pd.

Library¶

A library is a collection of reusable code. Examples used in EDA include pandas, NumPy, Matplotlib, and Seaborn.

Package¶

A package is an installable collection of Python modules. Packages are usually installed into a Python environment.

Module¶

A module is a Python file or package component that contains code. Modules may contain functions, classes, constants, and other definitions.

Local Variable¶

A local variable is a name you choose that refers to a value or object. For example, df often refers to a pandas DataFrame.

Object¶

An object is a value with data and behavior. In pandas, a DataFrame is an object with rows, columns, methods, and attributes.

Function¶

A function is reusable code that performs a task. For example, print() displays text, and len() returns the length of something.

Method¶

A method is a function attached to an object. For example, df.head() calls the head() method on the DataFrame named df.

Attribute¶

An attribute is a value stored on an object. For example, df.shape is an attribute that gives the number of rows and columns.

Argument¶

An argument is a value passed into a function or method. For example, in df.head(10), the value 10 is an argument.

Keyword Argument¶

A keyword argument is an argument passed by name. For example, in df.dropna(subset=cols_required), subset is a keyword argument.

List¶

A list is an ordered collection of values. For example, a list of column names may look like ["bill_length_mm", "body_mass_g"].

Tuple¶

A tuple is an ordered collection of values that is usually treated as fixed. For example, df.shape returns a tuple like (344, 7).

String¶

A string is text data. Column names, labels, file paths, and category names are often strings.

Integer¶

An integer is a whole number. Row counts and column counts are usually integers.

Float¶

A float is a number with a decimal point. Measurements such as body mass or bill length may be stored as floats.

Boolean¶

A Boolean is a true-or-false value. Boolean values are often used for filtering data.

Section 4. Using the pandas Library¶

pandas library (critical for analysts)¶

pandas is a Python library for working with structured data. It is especially useful for tabular data with rows and columns.

DataFrame¶

A DataFrame is a two-dimensional pandas table. It has rows, columns, column names, and an index.

df¶

df is a common variable name for a DataFrame.

Series¶

A Series is a one-dimensional pandas object. A single DataFrame column is usually a Series.

Index¶

The index labels the rows of a DataFrame or Series. The default index is often a sequence of numbers starting at 0.

Column Name¶

A column name is the label for a column in a DataFrame. Column names are used to select, filter, group, and summarize data.

dtype¶

dtype means data type. In pandas, each column has a dtype such as int64, float64, object, string, or category.

shape¶

shape is a DataFrame attribute that returns the number of rows and columns. For example, (344, 7) means 344 rows and 7 columns.

head()¶

head() shows the first few rows of a DataFrame. It is often used at the beginning of EDA to preview the data.

tail()¶

tail() shows the last few rows of a DataFrame. It can help confirm how the dataset ends.

info()¶

info() prints column names, data types, non-null counts, and memory usage. It is useful for checking structure and missing values.

describe()¶

describe() calculates summary statistics for numeric columns by default. It commonly returns count, mean, standard deviation, minimum, quartiles, and maximum.

isna()¶

isna() checks whether values are missing. It returns true for missing values and false for present values.

isnull()¶

isnull() is another pandas method for checking missing values. In pandas, isnull() and isna() are commonly used for the same purpose.

notna()¶

notna() checks whether values are present. It returns true for non-missing values.

sum()¶

sum() adds values. When used with Boolean true-or-false values, true values are counted as 1 and false values as 0.

mean()¶

mean() calculates the arithmetic average. When used on Boolean values, it can calculate the proportion of true values.

duplicated()¶

duplicated() identifies duplicate rows. It returns true for rows that repeat earlier rows.

dropna()¶

dropna() removes rows or columns with missing values. Using subset= limits the missing-value check to selected columns.

copy()¶

copy() creates a separate copy of a DataFrame or Series. It is often used to avoid accidentally modifying the original data.

select_dtypes()¶

select_dtypes() selects columns by data type. For example, it can select only numeric columns.

groupby()¶

groupby() splits data into groups based on one or more columns. It is commonly followed by an aggregation method.

agg()¶

agg() applies one or more aggregation functions to grouped data. For example, it can compute count, mean, standard deviation, minimum, and maximum by group.

corr()¶

corr() calculates correlations among numeric columns. It is commonly used to create a correlation matrix.

value_counts()¶

value_counts() counts how often each unique value appears. It is useful for categorical columns.

sort_values()¶

sort_values() sorts a DataFrame or Series by values. It can sort in ascending or descending order.

subset¶

A subset is a selected part of a larger dataset. A subset may include selected rows, selected columns, or both.

Filter¶

A filter selects rows that meet a condition. For example, a filter might keep only rows where species is Gentoo.

Chaining¶

Chaining means calling multiple methods in sequence. For example, df.isna().sum().sort_values() chains several operations.

Section 5. Visualization Terms¶

Visualization¶

A visualization is a graphical representation of data. Charts help reveal patterns, comparisons, distributions, and relationships.

Chart¶

A chart is a visual display of data. Common chart types include bar charts, histograms, scatter plots, box plots, and heatmaps.

Figure¶

A figure is the full Matplotlib drawing area. A figure may contain one or more plots.

Axes¶

Axes are the plot area where data is drawn. In Matplotlib and Seaborn, an Axes object is often used to set titles, labels, and formatting.

Bar Chart¶

A bar chart compares values across categories. It is useful for counts, averages, or totals by group.

Histogram¶

A histogram shows the distribution of a numeric variable. It groups values into bins and shows how many observations fall into each bin.

Scatter Plot¶

A scatter plot shows the relationship between two numeric variables. Each point represents one observation.

Box Plot¶

A box plot summarizes the distribution of a numeric variable. It shows the median, quartiles, spread, and possible outliers.

Heatmap¶

A heatmap uses color intensity to represent values in a table. Correlation matrices are often visualized as heatmaps.

x-axis¶

The x-axis is the horizontal axis of a chart. It often shows categories or an independent numeric variable.

y-axis¶

The y-axis is the vertical axis of a chart. It often shows counts, measurements, or a dependent numeric variable.

Label¶

A label describes an axis, category, point, or legend item. Good labels make charts easier to understand.

Title¶

A title explains what a chart shows. A good chart title should be specific enough to stand alone.

Legend¶

A legend explains visual encodings such as color, marker style, or group. For example, species may be shown using different colors.

Color Encoding¶

Color encoding uses color to represent categories or values. In EDA, color is often used to distinguish groups.

Section 6. Python Libraries Used¶

• NumPy is a Python library for numeric computing. It provides arrays and mathematical functions used by many data science libraries.
• pandas is used for reading, cleaning, transforming, summarizing, and analyzing structured data. Its primary data structures are DataFrame and Series.
• Matplotlib is a Python visualization library. It provides detailed control over figures, axes, labels, titles, and chart layout.
• Seaborn is a statistical visualization library built on top of Matplotlib. It provides convenient functions for charts such as scatter plots, histograms, box plots, and heatmaps.
• datafun_toolkit is a course-support package used for shared utilities such as logging. It provides functions such as get_logger() and log_header().

Section 7. Logging and Output¶

Logger¶

A logger records messages from code. In notebooks, a logger can show progress, data checks, warnings, and debugging details.

Log Message¶

A log message is a recorded message from the program. Examples include "Loading dataset" or "Duplicate rows detected: 0".

Log Level¶

A log level indicates the importance or purpose of a message. Common levels include DEBUG, INFO, WARNING, ERROR, and CRITICAL.

• DEBUG messages provide detailed information for troubleshooting. They are useful during development but may be too detailed for final reports.

• INFO messages describe normal progress. They are useful for showing what the notebook is doing.

• WARNING messages indicate something unexpected or potentially problematic. A warning does not always mean the notebook must stop.

• ERROR messages indicate that something failed. An error may stop a cell from completing.

print()¶

print() displays text output. It is simple and useful, but less structured than logging.

display()¶

display() renders rich output in a notebook. It is useful for showing DataFrames as formatted tables.

Section 8. Professional Communication¶

Standard Header¶

A standard header gives the notebook title, author, repository, purpose, date, and dataset information. It helps readers understand what the notebook is and where it belongs.

Purpose¶

The purpose explains why the notebook exists. A clear purpose helps readers understand what the notebook is meant to demonstrate or investigate. The ability to understand and covey your purpose for the exploration is critical.

Repository Link¶

A repository link points to the project source. It helps readers find related files, instructions, and version history.

Dataset Source¶

The dataset source explains where the data came from. This is important for credibility, reproducibility, and citation. The ability to document a datasource is critical.

A citation gives credit to the data creators or source. Datasets should be cited when the source provides citation information.

Clean vs Stale Output¶

Clean output means the notebook output is relevant, readable, and not cluttered with stale or unnecessary results. Before sharing a notebook, verify everything appears correctly when displayed in your GitHub repository.

Stale output is output that no longer matches the current code. This can happen when code is edited after cells have already been run.

Save Notebook¶

Saving a notebook writes the current code, Markdown, and possibly outputs to the .ipynb file. Always save after important changes.

Commit¶

A commit records a saved version of files in Git. A good commit captures a meaningful checkpoint in the project.

Section 9. Common EDA Workflow Terms¶

Load¶

Loading means bringing data into the notebook. For example, a dataset may be loaded into a pandas DataFrame.

Inspect¶

Inspecting means looking at the structure and contents of the data. Common inspection methods include head(), info(), shape, and columns.

Clean¶

Cleaning means preparing data for analysis. This may include handling missing values, fixing types, renaming columns, or removing duplicates.

Transform¶

Transforming means changing data into a more useful form. Examples include creating new columns, filtering rows, or grouping data.

Visualize¶

Visualizing means creating charts from data. Charts help identify patterns that may not be obvious from tables alone.

Interpret¶

Interpreting means explaining what the analysis suggests. Interpretation should stay grounded in the evidence shown by the data.

Finding¶

A finding is an observation supported by the analysis. Findings should be specific and connected to evidence from tables, statistics, or charts.

Summarize¶

Summarizing means reducing data into useful statistics or tables. Examples include counts, means, grouped summaries, and correlation matrices.

Limitation¶

A limitation is something the data or analysis cannot fully answer. Good EDA notes limitations clearly instead of overstating conclusions.

Next Steps¶

A next step is a reasonable action after EDA. Examples include additional cleaning, collecting more data, modeling, or investigating a pattern more closely.