STA261 Lecture1 Introduction
Uploaded by
garyclearwater862STA261 Lecture1 Introduction
Uploaded by
garyclearwater862STA 261: Probability & Statistics II
Introduction
Prof. Jesse Gronsbell
University of Toronto
Winter 2026
Welcome!
⋆ This is a continuation of STA257
⋆ We’ll move from probability to statistical inference
2 / 21
The goal of statistical inference
Learning about what we do not observe (parameters)
using what we observe (data)
3 / 21
Classic example: Is your lucky coin fair?
Question
How could you tell in a statistically rigorous way?
4 / 21
Classic example: Is your lucky coin fair?
Parameter: p = the probability of heads
Data: Toss your lucky coin 500 times and record heads or tails
Inference: Use your data to conduct a hypothesis test of
H0 : p = 0.5
HA : p ̸= 0.5
5 / 21
Classic example: Is your lucky coin fair?
Parameter: p = the probability of heads
Data: Toss your lucky coin 500 times and record heads or tails
Inference: Use your data to conduct a hypothesis test of
H0 : p = 0.5
HA : p ̸= 0.5
But coins aren’t very interesting.
And maybe coins aren’t quite fair
5 / 21
Statistical inference in the wild
Image Credit
Make a decision about whether to use Option A/B for an advertisement
based on the observed clickthrough rates.
6 / 21
Statistical inference in the wild
Slides by Lihua Lei
For a new test point, (X , Y = ?), find an interval, Ĉ (x), such that
P Y ∈ Ĉ (X ) ≥ 0.90.
7 / 21
Statistical inference in the wild
Seyyed-Kalantari et al 2021
Identify a disparity in the accuracy of
a chest X-ray classifier across race.
8 / 21
Statistical inference in the wild
Angelopoulos et al 2023
Use AI-generated data for valid and precise scientific discovery.
9 / 21
What is common to these examples?
10 / 21
What is common to these examples?
Decision making in the presence of uncertainty
11 / 21
Principle of statistical inference
Data is a realization of a random process
Why? The world isn’t deterministic, there is randomness:
⋆ In the study design (e.g., A/B test)
⋆ In the phenomenon (e.g., coin flips)
⋆ Due to sampling of the data (e.g., MIMIC-CXR)
⋆ Within the statistical model (e.g., ML model)
12 / 21
This class: Formalizing notions of uncertainty
⋆ Part I: A 100m dash through Probability Theory
Mathematical framework to describe data generating processes
⋆ Part II: An 800m race through Statistical Inference
Mathematical formalization of statements made about a data
generating process based on observed data
13 / 21
Probability = Inference−1
Wasserman 2004
14 / 21
Course Format
⋆ Most lectures will be around 2.5 hours
⋆ Office hours during time following the lecture
⋆ Tutorials start next week
⋆ Everything is on the course website:
[Link]
Make sure to carefully read the syllabus!
15 / 21
Assessments
⋆ 3 term tests (35%, 35%, 30%)
⋆ First term test (week of Jan 26)
⋆ See syllabus for the tentative schedule
16 / 21
Teaching assistants
⋆ Jianhu Gao
⋆ Madhu Gunasingam
⋆ Dayi Li
⋆ Konstantinos Tsiolis
⋆ Yuan Tian (lead TA)
⋆ Alexander Valencia
Feel free to attend any tutorial.
17 / 21
Office Hours
Schedule is TBD and will be posted on the course website.
OH for later this week will be announced via Quercus.
18 / 21
Communication
⋆ Piazza for anything related to content or logistics - not email!
⋆ Private matters can be sent to sta261@[Link]
⋆ Do not email me directly - you will not get a response
19 / 21
Computing
⋆ You will see R code in class
⋆ Instructions on how to download R are on the syllabus
20 / 21
Tips to do well
⋆ Come to class and tutorial
⋆ Read the book
⋆ Do practice problems
⋆ Ask questions - no question is silly
21 / 21