Contents

5 parts and 2 appendices

The first edition, from 2023, covers the foundations, exploratory analysis, inferential analysis and models. The second adds 10 chapters on data analysis in the age of AI.

To begin

Preface

  1. PrefaceHow this book began at UTN Haedo, what changes in the second edition, how to read it and under which license it can be shared and reused.

Part I

Foundations

Chapters 1 to 4

  1. 1Concepts and definitionsWhat data analysis is, how it relates to data engineering and data science, and what knowledge and tools the role requires.
  2. 2Working environmentUpdated 2026Spreadsheets, visualization tools, notebooks and cloud environments: how to choose where to work and what properties an analysis should have.
  3. 3Programming languagesPython, R, Julia, JavaScript and SQL in data analysis: what each language is used for and why they matter for growing in the role.
  4. 4Data, units and variables of analysisWhat a datum is, how it arises from crossing units of analysis with variables, and which types of variables and measurement scales exist.

Part II

Exploratory analysis

Chapters 5 to 7

  1. 5Frequency tablesThe most basic way to summarize data: absolute, relative and cumulative frequencies, and how to group quantitative variables into intervals.
  2. 6Describing variablesMeasures of location, dispersion and association: the three essentials for characterizing and summarizing variables in exploratory analysis.
  3. 7Data visualizationHow to choose a chart by the number and type of variables and the question at hand, and why the form depends on who receives it.

Part III

Inferential analysis

Chapters 8 to 15

  1. 8Data sources and collectionCensuses and samples: why where data comes from and how it is collected set the possible conclusions, and the role of randomness.
  2. 9Data wranglingUpdated 2026Joining, transforming, aggregating and filtering: the wrangling that takes up much of an analyst's work, and how to approach it in order.
  3. 10Applied probabilityThe basics of probability and the binomial, normal and chi-square distributions, the mathematical ground of inference.
  4. 11Point and interval estimationParameters, estimators and confidence intervals: how to approximate unknown features of a population from a sample.
  5. 12Hypothesis testsExtendedHow to test an estimator against an assumption about the true value, why it is science's tool of choice, and how to pick the test.
  6. 13Nonparametric statisticsGoodness-of-fit, independence and randomness tests that do not depend on parameters, and the bootstrap as a resampling method.
  7. 14Sampling techniquesSimple random, systematic, stratified and cluster sampling: how to get representative samples under real constraints.
  8. 15ANOVA: analysis of varianceHow to compare the means of more than two groups or treatments by analyzing variation within and between groups, with blocking and post hoc tests.

Part IV

Models

Chapters 16 to 18

  1. 16Linear regressionThe most widespread model for estimating a continuous variable: parameters, residuals, common data problems and its limits.
  2. 17Logistic regressionExtendedAn interpretable binary classification model: probabilities, the cutoff point and the confusion matrix for evaluating its predictions.
  3. 18KNN and CARTExtendedNearest neighbors and decision trees: flexible models for regression and classification, useful for exploring and as baselines.

Part V

Data analysis in the age of AI

Chapters 19 to 28

New in the second edition
  1. 19From copilot to colleagueAI tools went from suggesting code to chaining whole analyses. What changes in the analyst's work and what remains theirs.
  2. 20Asking the data in natural languageQuerying data in words instead of SQL: how it works, why ambiguity is the underlying problem and how to verify each query.
  3. 21Verifying what the machine saysHallucinations, variability and evaluation sets: how to measure how far to trust an AI tool with the tools of statistics.
  4. 22Agents that analyzeA local warehouse per agent and a central one for the team: changes with an author, a key and an owner per table, conflicts set aside, never overwritten.
  5. 23Every number with its sourceA report made by agents states the data version and cites the table, row or query behind each figure, with no unsupported claims.
  6. 24Validate before you believeWhat research measures about validating agents' work, how to read those figures with intervals, and why validation is a hypothesis test.
  7. 25Text, images and audio as variablesHow to turn comments, photos and recordings into analyzable variables with AI, validating the measurement and reporting its margin of error.
  8. 26Synthetic data and simulationWhat model-generated data is good for, what it cannot replace, the risk of model collapse and how to evaluate a synthetic set.
  9. 27Privacy, bias and responsibilityPersonal data, sending data to external tools, biases born in the sample, and why responsibility for an analysis is never delegated.
  10. 28From insight to actionFrom conclusion to automated decision: uncertainty, cutoffs, controlled experiments, monitoring and people in the loop.

Appendices

Appendices

Appendices A and B

  1. AProgramming fundamentalsData types and structures, operators, control structures and functions: the concepts common to any programming language.
  2. BCopilots and chatbotsWritten in February 2023: how the first copilots and ChatGPT were changing how data analysis was learned and practiced.

About this edition

Two editions, one book

The first edition was written in 2023 as course material for the University Diploma in Programming at UTN Haedo. Each chapter opens with a summary of the key ideas and goes on to further reading, reviewed for this edition.

The second edition keeps that material and adds 10 chapters on how working with data changes when AI is one more tool: delegating tasks, asking in natural language, verifying results, measuring what is not in a table, simulating, protecting privacy and moving to action. Always connected to the foundations of the first.

License. Data Analytics, by Ervin Wilneder, is published under the Creative Commons Attribution 4.0 International license (CC BY 4.0): it may be copied, shared and adapted, even commercially, as long as the source is credited.

Frequently asked questions

About Data Analytics

What is Data Analytics?
Data Analytics is an open book by Ervin Wilneder on concepts, tools and techniques, from statistics to the age of AI. Second edition, 2026, under the CC BY 4.0 license. An open book on data analysis: descriptive and inferential statistics, models, and a second edition on working with data in the age of AI.
Who wrote Data Analytics?
Ervin Wilneder. The first edition is from 2023 and the second, from 2026. This is the English edition; the original is in Spanish.
Can I read it for free?
Yes. The whole book can be read at ervinwilneder.ai, free and without signing up, and downloaded as PDF and EPUB.
Can I share or reuse it?
Yes. It is published under the Creative Commons Attribution 4.0 International license (CC BY 4.0): it may be copied, shared and adapted, even commercially, as long as the source is credited: Data Analytics, by Ervin Wilneder.
How is it organized?
In 5 parts (Foundations, Exploratory analysis, Inferential analysis, Models and Data analysis in the age of AI), with 28 chapters and 2 appendices.