noBS STATS Notebooks

Hands-on tutorials, exercises, and projects to accompany the No Bullshit Guide to Statistics. Motivation: reading about stats is not enough... you need to experiment with stats, play with different examples, and do some real-world data analysis tasks. This is what this repo is about.

Getting started

Use the binder button below to start an ephemeral JupyterLab instance where you can run the code in each notebook.

Contents

Once have launched JupyterLab instance, use the file browser (folder icon in the left pane) to navigate to any of these subfolders and try the notebooks in them:

• datasets/: all the data used for examples, exercises, and problems in the book.
• notebooks/: notebooks to accompany each section of the book.
• stats_overview/: a complete worked example to introduce all the statistics concepts from the book: DATA, PROB, STATS, and LINEAR MODELS.
• exercises/: simple, routine exercises to practice new concepts.
• problems/: problems requiring some thinking COMING SOON
• missions/: multi-step procedures and workflows COMING SOON
• tutorials/: tutorials that introduce Python basics, and the Pandas and Seaborn Python libraries.

Other stuff

This repo also contains the "utility" code and notebooks that were used for the book:

• data_generation: notebooks used to generate datasets used in the book.
• figures_generation: notebooks to run to generate figures and tables.
• notebooks/explorations/: general exploratory notebooks on statistics.
• notebooks/drafts/: notebook drafts (not ready for prime time).

Tech stack

• Python and notebooks runnable in jupter-lab
• data manipulation using pandas and some numpy
• graphing with seaborn
• prob. and stats functionality from scipy
• advanced stats from statsmodels

All of these can be installed by running pip install -r reqirements.txt in the root of this project.

Pedagogical notes

• The goal is to have lots of hands-on activities, called missions, that ask readers to do a complete statistical analyses (data->model->stats->report)
• Each mission will provide some scaffolding so learners are not left completely on their own (for beginners 3/4 of the steps are already filled in, then for intermediate learners notebook is half-filled, and advanced learners are just provided with the goal and otherwise left to work independently)
• Most common task: write a Python function that fulfills a detailed specification, including what inputs to accept and what output the function is expected to produce.
• Student's solution (a filled-in function) gets tested using a test function, just like unit test used in software projects.
  • Prior work: tutorial ask learners to assign their answer to a variable: assert statements We'll be taking this idea to the next level by asking learners to define a callable.

TODOs

• write CI script to strip out exercise solutions from notebooks