Feedback on Linear Regression Assignment

Be curious!

LLMs!

Carlijn & Georgia

Commits
- How many Carlijns?
- Georgia one commit?
- Commit messages good but be a bit more specific
- Cartoon on informative git commit messages
Good explanatory text and layout.
Good residual and scale location plots. But what can we do about the nonlinearity and heteroscadisticity?
Correlations: what steps to take?
Think about your layout in the answers to 9c
9d scale location and influence plots are not the prettiest. How could you make the plot more readable?
"…exclusion from the model would likely result in a noticeable change in the estimated coefficients."

Emil, Adrian, Louanne

No report, just code
Where to look? Various files.
Comments in Python files

Spider file headers

# -*- coding: utf-8 -*-
        """
        Created on Thu Sep 25 12:19:18 2025

        @author: seinc
        """

You might consider a separate file for imports and setup.

Matúš, Jesse, Nora

Single repo. We will use one repo per assignment.
Lab1 in Python file. Clear code with comments.
Lab2 files not obvious to find.
Clearly laid out. Good formatting
"High multicollinearity may require careful handling in regression"
"significant contributors to predicting mpg"
QQ-plots good!
Lots of plots!
Commits
Watch out for how some editors save your files
.gitignore and .DS_Store in this commit

Madalina, Natalia

Week 1
- All commits by Natalia
- Learning goals and questions stated in report.
- Code and answers in Python
- No plots

Yaohong, Natalia

Many commits by Natalia
Short direct answers in the beginning. Better explanations later.

Roman, Lynn

lviss
Lots of Roman
Commit messages
Put yourself into the reader's shoes.
Images of text!

Janina and Mariana

Most balanced commit history!
Images subdir
Step by step explanation
Good formatting. Drawing attention to results in bold.
Don't let plots get too "busy". See Tufte
Something went wrong with the formatting.

Max

Commented code for exercises.

Nora and Jagoda

Balanced commits with encouragement!
All in single Python file with comments
Zone Identifier files

Fridolin & Efraim

Plots with little explanation

Selecting many columns

desc_stat = auto.select(
    pl.col("mpg"),
    pl.col("cylinders"),
    pl.col("displacement"),
    pl.col("horsepower"),
    pl.col("weight"),
    pl.col("acceleration"),
    pl.col("year"),
).describe()

desc_stat = auto.select(
    [
        "mpg",
        "cylinders",
        "displacement",
        "horsepower",
        "weight",
        "acceleration",
        "year",
    ],
).describe()

Same applies to savefig as to reading data. Better use relative dir

plot_9_d.savefig(
    "/Users/efraimtoth/Library/CloudStorage/OneDrive-UvA/AUC/3rd year/Machine Learning/repos/lab1/Images/auto_pairplot.png"
)

Efraim & Romane

Commits in shifts
Image sizing
Table formatting

Yaohong & Emil

In Python files with comments
Each one did a complete lab.