Sources and further study

The explanations and activities in this edition are original teaching material built around the author’s retained book. Sources below support particular definitions, software behaviour and data provenance. The private course library was used for coverage comparison; its prose, solution notebooks and branding are not redistributed as this book’s content. Documentation was inspected on 24 October 2026; the environment record identifies the versions actually executed.

Data and scientific foundations

• Cleveland Museum of Art. Open Access API and Open Access terms. Metadata for the cultural and text projects; saved query and hashes are supplied.
• Horst, A. M., Hill, A. P., and Gorman, K. B. (2020). palmerpenguins. Simplified measured penguin data collected by Gorman and Palmer Station LTER; source attribution and CC0 terms are retained.
• World Bank. World Development Indicators, SP.POP.TOTL metadata, API call structures, and data licences. Selected country-year snapshot used in the capstone.
• Frické, M. (2019). The Knowledge Pyramid: the DIKW Hierarchy. Knowledge Organization, 46(1), 33–46. doi:10.5771/0943-7444-2019-1-33. Discusses the model and substantive criticisms; it is not cited as proof that data automatically turn into wisdom.
• Hernán, M. A., and Robins, J. M. (2020). Causal Inference: What If. Chapman & Hall/CRC. Authors’ book page. Further study of intervention-based causal questions beyond descriptive comparisons.
• American Statistical Association (2016). Statement announcement on statistical significance and p-values. Supports the cautions about reducing scientific interpretation to a significance threshold.
• NIST. Correlation reference. Mathematical background for linear correlation.

Numerical and tabular interfaces

• NumPy Developers. Copies and views, broadcasting, Generator, standard deviation, and least squares.
• Pandas Developers. Indexing, advanced indexing, I/O, missing data, text, merging, reshaping, time series, and scaling.
• SciPy Developers. Bootstrap reference. Further methodological context; the small resampling experiment uses NumPy and does not require SciPy.

Diagnostic figures and preparation for later books

• Matplotlib Developers. Quick-start guide and box-plot reference. Figure/Axes interface and default box-plot convention.
• scikit-learn Developers. Evaluation, cross-validation, and common pitfalls. Metric definitions, evaluation boundaries and preprocessing leakage.
• Hugging Face. Dataset cards. Provenance and dataset documentation; authenticated acquisition is not required for this book.
• Kaggle. API documentation. Optional acquisition reference; the book does not distribute credentials or require an authenticated Kaggle download.

A complementary book

McKinney, W. (2022). Python for Data Analysis, third edition. O’Reilly. Open edition. Its coverage of arrays, tables, wrangling, time and analysis examples was inspected as a complementary reference. This edition’s examples, project questions and solutions were written independently, not reproduced from that book.

References used in the scientific accuracy review

• NumPy Developers. Universal functions, NumPy 2.1. Distinguishes ufunc interfaces from other numerical routines.
• NumPy Developers. Normal generator. Defines the simulated normal model; it does not validate a model of real climate measurements.
• Pandas Developers. GroupBy minimum, count, and idxmin. Column reductions, non-missing counts and first-minimum row labels.
• scikit-learn Developers. Imputation of missing values. Further explanation of imputation methods and alternatives; the variance counterexample in this book is derived directly from its supplied values.

These sources were inspected on 24 October 2026. The corrected least-squares coefficients, imputation variance, selected employee and book totals are independently checked arithmetic for invented teaching inputs, not empirical findings about real people or institutions.