Introduction to Statistics and Data Science

Goals of the Course

Statistics and data science have traditionally been taught through theoretical lectures, textbook readings, and problem sets. While this approach has served generations of students well, it leaves two persistent gaps. The first is a technical barrier: before a single regression can be estimated, students must install statistical software, navigate unfamiliar syntax, manage data files, and troubleshoot installation problems, all of which diverts energy away from the concepts themselves. The second is the gap between "knowing" and "doing": a student may follow the algebra an statiatical estimator and still be unable to translate that knowledge into working code that analyzes real data.

This course is designed to close both gaps. It combines the rigor of a traditional introduction to statistics with cloud-based computational notebooks and AI-assisted learning tools. The foundational content follows A. Colin Cameron's "Analysis of Economics Data: An Introduction to Econometrics" (2022), and every concept is implemented in Python notebooks that run directly in the browser through Google Colab, with zero installation required. Students begin working with real economic and development data from the very first session, and they progress systematically from descriptive statistics, through the sample mean and statistical inference, to bivariate and multiple regression, model diagnostics, and finally to panel data, time series, and the modern research designs used to establish causation. The aim is to turn what is often experienced as a daunting technical subject into an engaging process of discovering economic stories in real data.

Objectives of the Course

This course enables students to move fluently between statistical reasoning and its computational implementation, so that they can use data to answer their own research questions rather than only follow the analyses of others.

Upon completion of this course:

• Students will be able to summarize, transform, and visualize economic and development data, and to describe a variable's distribution and the relationship between two variables using appropriate statistics and charts.

• Students will be able to estimate, interpret, and test bivariate and multiple regression models; construct confidence intervals; conduct single and joint hypothesis tests; use heteroskedasticity-robust and cluster-robust standard errors; and diagnose problems such as multicollinearity, heteroskedasticity, autocorrelation, outliers, and influential observations.

• Students will be able to distinguish correlation from causation, and to read and critically evaluate empirical evidence based on randomized controlled trials, difference-in-differences, regression discontinuity, instrumental variables, and panel data methods.

• Students will be able to write, run, and document a reproducible Python notebook in Google Colab that takes a dataset from raw form to estimated results and a written interpretation, and to use AI learning tools responsibly, verifying their output against authoritative sources rather than accepting it at face value.

Teaching tips and student responsibilities

THREE PILLARS OF LEARNING. This course is built on three complementary pillars that work together as a single learning ecosystem. The first pillar, foundational concepts, keeps the statistical and econometric content rigorous by following Cameron's (2022) pedagogical framework. The second pillar, computational notebooks, gives every chapter a companion Google Colab notebook so that students learn by coding from the very first session, with no software installation. The third pillar, AI-powered learning, surrounds each topic with visual summaries, short animated video overviews, presentation slides, podcast episodes, self-assessment quizzes, and an AI tutor, so that students can approach the same material through whichever modality suits them best. Students are encouraged to use all three pillars rather than only the one they find most comfortable.

ZERO INSTALLATION. There is nothing to install and nothing to configure. Every notebook opens in the browser with a single click on an "Open in Colab" badge, and it arrives with all libraries, datasets, and code already in place. Students can work from a Windows PC, a Mac, a Chromebook, or a tablet, and can start an analysis at home and continue it on campus, because everything is saved in the cloud. Removing the setup barrier means that class time and study time are spent on statistics rather than on troubleshooting.

BEFORE EACH SESSION. Preparation is what makes the classroom experience meaningful, and it is deliberately kept light. Before each session, students should read the corresponding chapter of Cameron's textbook for the theoretical context and the mathematical derivations, watch the short video overview (under ten minutes) to build a mental framework for the concepts that are coming, and skim the visual summary to see how the chapter's ideas fit together. Arriving with a rough map of the territory allows the session itself to move faster and go deeper.

IN THE CLASSROOM. Sessions are run as live coding sessions rather than as one-way lectures. Concepts are introduced, then immediately estimated on real data, with students running and modifying the code cells on their own screens as the class proceeds. Interactive learning apps are used to build intuition before the algebra: students drag a sample-size slider and watch a sampling distribution narrow in real time, toggle between model specifications and see R-squared change instantly, or launch a Monte Carlo simulation with one click and watch unbiasedness emerge. Because the pace is brisk, students are expected to engage actively, ask questions as soon as something is unclear, and treat every result on screen as something to be interrogated rather than copied.

OUTSIDE THE CLASSROOM. Students should plan on two or more hours of independent study for every class hour. The most important habit is to run each notebook cell by cell rather than passively reading it: study the code, make sure each section's purpose is clear, then experiment by changing a parameter, substituting a different dataset, or extending an analysis beyond what is shown. Deep learning happens in that experimentation. Afterwards, the chapter dashboard is useful for consolidating intuition, the podcast episode works well for review while commuting or exercising, and the quiz should be taken seriously as a diagnostic tool: it is far better to discover a gap in understanding through a quiz than through an exam. Problem sets should be worked through with attention to the solution process, not only to the final answer. When students get stuck, the AI tutor is available at any hour to explain code, clarify a theoretical point, or help debug an error, so no one needs to remain blocked.

READING MATERIALS. Cameron's textbook explains the "why" behind each method, while the notebooks show the "how"; the two are meant to be read together. A pre-reading pass before the lecture makes comprehension in class considerably faster, and a careful post-lecture review of the notebook, the slides, and one's own notes is what turns temporary familiarity into durable understanding. Cameron's original slides are also provided alongside the AI-generated slides, so that students can compare the authoritative instructor presentation with an alternative explanation of the same material.

RESPONSIBLE USE OF AI. AI tools are used intensively in this course, and their proper role is stated explicitly: they are an enhancement, not a replacement for critical thinking and genuine understanding. The foundation of the course remains Cameron's textbook and the verified Python code in the notebooks. AI-generated explanations can occasionally be mistaken or can oversimplify a subtle point, so students are expected to cross-reference them against these authoritative sources and against the tested code. The responsibility for verifying a result and for being able to explain it in one's own words always rests with the student. Used this way, AI tools open multiple pathways to the same understanding; used uncritically, they substitute for it. Learning the difference is itself one of the skills this course intends to teach.

Course outline

The course is delivered in fifteen weekly sessions. It covers the full sequence of a first course in statistics and econometrics, organized in four parts, and follows the seventeen chapters of the companion book, with two pairs of chapters taught together.

PART I. STATISTICAL FOUNDATIONS

1. Analysis of economics data and univariate data summary (Chapters 1-2)
2. The sample mean and sampling distributions (Chapter 3)
3. Statistical inference for the mean (Chapter 4)

PART II. BIVARIATE REGRESSION

4. Bivariate data summary (Chapter 5)
5. The least squares estimator (Chapter 6)
6. Statistical inference for bivariate regression (Chapter 7)
7. Case studies for bivariate regression and models with natural logarithms (Chapters 8-9)

[ Midterm examination covers Sessions 1-7 ]

PART III. MULTIPLE REGRESSION

8. Data summary for multiple regression (Chapter 10)
9. Statistical inference for multiple regression (Chapter 11)
10. Further topics in multiple regression (Chapter 12)
11. Case studies for multiple regression and causal designs (Chapter 13)

PART IV. FURTHER TOPICS AND ADVANCED METHODS

12. Regression with indicator variables (Chapter 14)
13. Regression with transformed variables (Chapter 15)
14. Checking the model and the data (Chapter 16)
15. Panel data, time series data, and causation (Chapter 17)

[ Final examination in the examination period ]

Course Evaluation Method and Criteria

Midterm exam (30%); Final exam (40%); Problem sets and quizzes (30%)

The midterm examination covers Sessions 1 to 7. The final examination covers the whole course, with emphasis on Sessions 8 to 15. Problem sets are submitted as Google Colab notebooks and are assessed on the correctness of the analysis and on the quality of the written interpretation, not only on the final numbers. Chapter quizzes are taken online and provide immediate feedback.

Students are expected to devote two or more hours of independent study for every class hour.

Textbook/Reference Book

1．Textbook

Cameron, A. Colin (2022), Analysis of Economics Data: An Introduction to Econometrics. Companion materials, including datasets, detailed slides, and original Stata, R, and Gretl code, are openly available at: https://cameron.econ.ucdavis.edu/aed/index.html

2．Companion book and computational materials

Mendez, Carlos (2026), Econometrics Powered by AI: An Introduction Using Cloud-based Python Notebooks. Online book: https://quarcs-lab.github.io/metricsai/book/_book/
Companion website: https://quarcs-lab.github.io/metricsai

This companion book brings the lessons and examples of Cameron (2022) into an interactive Python environment. It provides one Google Colab notebook per chapter, an interactive learning app per chapter, and AI-generated study materials. The Python stack used throughout consists of pandas for data manipulation, PyFixest for regression estimation, inference, and reporting (with statsmodels and linearmodels for selected topics), matplotlib and seaborn for visualization, and NumPy and SciPy for numerical computing. All of it runs in Google Colab, so no local installation is needed.

3．General readings

No additional purchase is required. Supplementary readings and datasets drawn from economics and development research are distributed through the course site as each topic is introduced.

Lecture materials

All course materials are openly available online and require no installation. Each of the seventeen chapters of the companion book has its own set of the resources listed below, and all of them are reachable from the companion website at https://quarcs-lab.github.io/metricsai

• Interactive Google Colab notebooks, one per chapter, containing the full Python implementation of every concept, table, and figure, with commented code and step-by-step explanations.
• Interactive learning apps, one per chapter: self-contained browser dashboards with sliders, toggles, and real-time charts that let students explore key concepts without writing code. They use the same datasets as the notebooks, so the numbers match.
• AI-generated video overviews, one per chapter, each under ten minutes, explaining the chapter's key ideas through animation and narration.
• AI-generated visual summaries that condense each chapter into a single graphical map of its concepts and their relationships.
• AI-generated presentation slides for self-paced study and group discussion.
• Professor Cameron's original slides, datasets, and Stata, R, and Gretl code, at https://cameron.econ.ucdavis.edu/aed/index.html
• Online chapter quizzes with immediate feedback, and AI tutors available for code explanations and conceptual questions.
• Podcast episodes, one per chapter, presenting the content as a conversation, for review away from the computer.
• The full text of the companion book, at https://quarcs-lab.github.io/metricsai/book/_book/

This lecture is provided under Creative Commons Attribution-Non Commercial-ShareAlike 4.0 International.

---