0
! search... ! rnugroho Have a problem?
Menu
Remember that the quality of the defenses, hence the quality of the of the school on the labor market depends on you. The
My projects remote defences during the Covid crisis allows more flexibility so you can progress into your curriculum, but also brings
more risks of cheat, injustice, laziness, that will harm everyone's skills development. We do count on your maturity and
Holy Graph wisdom during these remote defenses for the benefits of the entire community.
List projects
Available Cursus
SCALE FOR PROJECT DSLR
Your projects
You should evaluate 1 student in this team
computorv1
ft_linear_regression
Git repository
Part_Time II git@vogsphere-v2.42.fr:vogsphere/intra-uuid-39015ea6-fce4-42cd-a954-0bfe0b7f11ea-
red-tetris
Introduction
For the smooth running of this evaluation, please respect the following rules:
- Remain polite, kind, respectful and constructive whatever happens during
this conversation. It's a matter of confidence between you and the
42 community.
- Highlight the potential problems you ‘ve had with the work you're presented
to the person or the group you're grading, and take the time to talk about
and discuss those issues.
- Accept the fact that the exam subject or required functions might lead
to different interpretations. Listen to your discussion partner's
perspective with an open mind (are they right or wrong ?) and grade them as
fairly as possible.
42's teaching methods can make sense only if peer-evaluation is
taken seriously.
Guidelines
- You must only evaluate what you will find in the student's or group's
GiT repository.
- Take the time to check that the GiT repository matches the student or
group and the project.
- Double check that no malicious alias was used to mislead you and make you
grade something different from the official repository content.
- If a script supposed to help evaluate the exam is supplied by either side, the
other side will have to strictly check it to avoid nasty surprises.
- If the evaluating student has not yet taken this project, they will have to
read the exam subject in its entirety before starting the evaluation.
- Use the flags available on this grading system to signal an empty or non-
funcional project, a norm flaw, cheating, etc. In that case, evaluation stops
and final grade is 0 (or -42 if it's a cheating problem). However, if it's
not a cheating problem, you are invited to keep talking about the work that
has been done (or not done, as a matter of fact) in order to identify the
issues that lead to this stalemate and avoid it next time.
Attachments
evaluate.py dataset_truth.csv
subject.pdf datasets.tgz
Data analysis
In this part, we will study the succinct data analysis through the 'describe' function.
The describe function
Execute the 'describe' function with 'dataset_train.csv' in parameter. Does
the output respect the requirement of the subject? That is: count, mean,
std, min, 25%, 50%, 75% and max.
Yes No
Hands in code
Open the 'describe' source and talk about the code together. Make sure the
assessed student doesn't use any third party library that would replace
one of the requested results. For instance: no 'mean' function prompting
the student would not have coded himself.
If the assessed student is using a prohibited function, check the Cheat
flag and end the evaluation. Validate only if they coded everything
themselves.
Yes No
Notions explanations
Ask the assessed student to explain the following notions:
- What is the average (mean)?
- What is the standard deviation (std)?
- What is a quartile (25% - 50% - 75%)?
1 correct answer = 1 point, 2 correct answers = 3 points, 3
correct answers = 5 points.
Rate it from 0 (failed) through 5 (excellent)
Data visualization
Here, we're going to tackle data visualization. This section will require a little thinking more than just development
skills. You will be the one to judge if the assessed student answers the question and if his explanations are satisfying.
If you're not satisfied with an answer, it might be wise to sit and think of another solution together. There might be
more than one anwser to a given question.
Histogram
Launch the `histogram` script.
Does the displayed graphic help you answer the question:
Which Hogwarts class has an homogenous grade repartition between the four
houses?
Ask the assessed student to explain what you see and why they believe it
answers the question.
Yes No
Scatter plot
Launch the `scatter_plot` script.
Does the displayed graphic help you answer the question:
which two features are similar?
Ask the assessed student to explain what you see and why they believe it
answers the question. For this part, there should only be one obvious
answer.
Yes No
Pair plot
Launch the `pair_plot` script.
Does the graphic help you answer the question:
from this graph, which characteristics will you use to train your coming
logistic regressions?
Ask the assessed student to explain what you see and why they believe it
answers the question.
Yes No
Logistic regression
We are going to evaluate the multi-classifier.
Discussions
Before launching any program, ask the assessed student how the logistic
regression works.
We're not here to nitpick but to make sure the assessed student has
understood the following points: how logistic regression works compared to
to linear regression, point in nornmalising the data, what's the one-vs-all
method. Of course, you can go further than these elements, but don't try
to push or trick the student.
Did the student give the correct explanations?
Yes No
Machine learning!
Time to evaluate the algorithme. First, execute `logreg_train` with
`dataset_train.csv`. This should create a file containing the weights for
each model. Is this the case?
Yes No
Predictions
Once you have trained your models, execute `logreg_predict` with the
weights and `dataset_test.csv`as parameters. This should create a file
named `houses.csv`.
In order to evaluate the multi-classifier performance, use the script
`evaluate.py` which will compare the files `houses.csv` with
`dataset_truth.csv` containing the truth (that is, the real houses
the students belong to).
Mc Gonagall had asked for a minimum score of 98% (equals 0.98). If this is
so, you can validate. Otherwise... Too bad.
Yes No
Bonus
Reminder: if, somehow, the program doesn't react as it should (bus error, segfault etc...), evaluation ends and the
grade is 0. Use the respective flags. This instruction works during the whole evaluation. Bonus will be taken into
account only if the mandatory part is PERFECT. PERFECT meaning it is completed, that its behavior cannot be
faulted, even because of the slightest mistake, improper use, etc... Practically, it means that if the mandatory part is
not validated, none of the bonus will be taken in consideration.
Let's talk, now.
Feel free to grade any additionnal features in the project. It will
remain at your discretion as long as you have good reasons to do so.
Rate it from 0 (failed) through 5 (excellent)
Ratings
Don’t forget to check the flag corresponding to the defense
Ok Outstanding project
Empty work W Invalid compilation Cheat d Crash l Forbidden function
Conclusion
Leave a comment on this evaluation
Finish evaluation
General term of use of the site Privacy policy Legal notices Declaration on the use of cookies Terms of use for video surveillance Rules of procedure
