Goals for Today (and next time)

Today:

Guess the age of Reed math / stats / cs faculty
Start to think about quantifying error

Friday (lab):

Technologies, Data, and Stat Learning Methods

Before we get started: Pre-lab assignment

Before Thursday 9/3 (tomorrow) at noon, please do the following:

We’ll use GitHub for our class, so each of you need to have a GitHub profile. If you don’t have a GitHub profile already, go to https://github.com/ and sign up for one. Once you’ve signed up for an account, please add it to the Google Sheet so that I can add you to the stat-243-f26 organization. Note: you’ll need to be logged in to your Reed Google account to edit the Google Sheet.

How Old Are We?

The Rules

I will display photos from 8 current and former math/stats faculty:

Goal: Estimate the age of each faculty member (at the time the picture was taken)
Note: None of these pictures were taken this year!

Work in groups of 2-3 to develop a consistent method for estimating age.

Record your estimates for each faculty member on paper and submit it to me.

Photo 1

Photo 2

Photo 3

Photo 4

Photo 5

Photo 6

Photo 7

Photo 8

All Photos

As we wait for groups to guess ages, complete the reflection worksheet.

Questions

Reflection: Modeling Questions

What explicit features are you analyzing?
How do you convert these features to numerical estimates?
How do you resolve differences in opinion among group members?

Reflection: Definitional Questions

Is the activity supervised or unsupervised?
- Supervised
Does it represent a classification or regression problem?
- Regression
Are you interested primarily in prediction or inference?
- Prediction
Are you using parametric or non-parametric methods?
- Either would have worked!

Reflection: Conceptual Questions

What are some sources of “error” in your estimates?
How can we quantify the error in your estimates?
How should we assess the overall accuracy of a group’s predictions?
In what other circumstances do people (or companies, or algorithms) predict age based on photos or physical appearance? Is that ethical?

Results

(on the board)

Returning to conceptual questions

What are some sources of “error” in your estimates?
How can we quantify the error in your estimates?
How should we assess the overall accuracy of a group’s predictions?
In what other circumstances do people (or companies, or algorithms) predict age based on photos or physical appearance? Is that ethical?

Next Time

Lab 1: Technologies, Data, and Stat Learning Methods

Remember, before Thursday 9/3 (tomorrow) at noon, please do the following: