CMSDAS Pre-Exercise 7: Statistics reading course
Overview
Teaching: 0 min
Exercises: 0 minQuestions
Which statistics concepts should I know before CMSDAS?
Where can I learn them, at my own pace?
Objectives
Learn the basics of each topic in the statistics reading course before CMSDAS.
Run the interactive examples to build intuition.
Optional, but a real head start
- This episode is not required to register for CMSDAS and has no questions to submit, but we strongly encourage everyone to do it.
- Almost every CMSDAS Long exercise ends in a fit, a significance, or a limit. Students who arrive knowing the basic concepts will have a real advantage, instead of learning the statistics under time pressure during the school.
- The same concepts are behind every CMS result you will work on after CMSDAS, so the effort pays off well beyond the school.
- Covering the basics takes steady effort over a few weeks, so start early.
The course
Statistical Methods and Data Analysis Techniques is an online book of reading material on the basic concepts of statistics.
Why this course?
- Written by a CMS physicist, in the language physicists use, and focused on the topics most relevant to HEP analyses.
- Interactive: many sections are Jupyter notebooks you can run and modify in your browser via Binder, with no installation needed.
Acknowledgement
The course is written by Mauro Donegà (ETH Zürich), based on the ETH Zürich master’s course Statistical Methods and Analysis Techniques in Experimental Physics, taught with Christoph Grab and Christian Regenfüs. We thank Mauro for allowing CMSDAS to use this material.
How to use it
Working through the full course at your own pace takes about 3 weeks. You don’t need all of it before CMSDAS: aim to learn the basics of each topic below, and go deeper wherever you are curious or where your analysis will need it.
- Read the book online in order, using the arrows at the bottom of each page, or jump to a topic from the sidebar.
- On notebook pages, click the rocket icon (top right) → Binder to run the code. Binder takes a minute or two to start, and its sessions are temporary: download the notebook (download icon) if you want to keep your changes.
| Topic | Read | Try it interactively |
|---|---|---|
| Probability | Introduction, Probability | Conditional probability, Bayes theorem |
| Distributions | Probability distributions | Random walk |
| Uncertainties | Measurement uncertainties | Error matrix |
| Monte Carlo | Monte Carlo methods | Random numbers with numpy |
| Parameter estimation | Statistical inference, Likelihood, Least squares | ML method: mean of a Gaussian, From least squares to likelihood |
| Hypothesis testing | Hypothesis testing | |
| Confidence intervals and limits | Confidence intervals | Poisson intervals, Upper limit |
| Multivariate analysis | Multivariate analysis methods |
To check your understanding, try the exercises at the end of the book.
Questions and discussion
There is a dedicated Mattermost channel for this course: StatisticsReadingCourse. Please ask questions there, however basic: someone else is probably wondering the same thing, and discussing with other students is a great way to learn. The channel is in the CMSDAS@LPC 2027 Mattermost team; if the link doesn’t open, join the team first (see the setup instructions).
Key Points
The reading course covers the statistics used in HEP analyses: probability, uncertainties, Monte Carlo, likelihood fits, hypothesis tests, and confidence intervals.
Many sections are executable notebooks you can run in your browser with Binder.
Ask questions in the StatisticsReadingCourse Mattermost channel.