search

rolling-scopes-school / support

15 stars 3 forks source link

Cross-Check '2 - Linear Regression and Visualization' - 'Kaktuscha' #189

Closed ghost closed 3 years ago

ghost commented 3 years ago

Repository: https://github.com/Kaktuscha/Linear_regressionn/blob/master/seminar_and_homework.ipynb Task: https://github.com/rolling-scopes-school/ml-intro/blob/2021/2_linear_regression/seminar_and_homework.ipynb Form: https://rolling-scopes-school.github.io/checklist/ Marks: Screenshot: image

4 students have checked out 5, and marks did not appear in the total score section.

JackNightsky commented 3 years ago

Проверив работу у меня получилась оценка в 95 балов. Но потом я глянул бал в Score таблице и там оказался 0 (ноль). Видется мне, что товарищ не проверял чужие работы. За что и получил свой 0. В изменении оценки смысла не вижу.

Средний бал вместе с моей из выставленных составляет (98 + 130 + 95 + 118) / 4 = 110.25 Такую оценку он мог бы получить.

TOTAL SCORE: 8 + 0 + 0.5 - 0.5 + 0.5 + 0.5 + 0.5 + 0 = 9.5 Kaktuscha

Task 1 (1 point) || OK Task 2 (0.5 point) || OK Task 3 (1.5 points)|| OK Task 4 (5 points) || OK

score: 8

Task 5 (up to 5 points)

  1. Extract float number from Value field in DataFrame (0.5 points) BAD 60K => 60 000 000 неверная конвертация тысяч. (60K == 60 000) || 0

  2. Сhoose more features that you expect to influence on player Value (at least 10) Plot feature correlation matrix. (0.5 points) OK || 0.5

  3. _Drop features that are highly correlated with each other (abs(corr) > 0.9) one by one until no correlated pairs left. Hint: you may reuse code from Task_9 in HW1 for automatic correlated pairs selection. (1.5 points) Penalties: -0.5 points if did not remove linearly dependent features before training the model. || -0.5

  4. Split data into train/test with some proportion (0.5 points) OK || 0.5

  5. Train a model on train dataset, make predictions both for train and test. (0.5 points) OK || 0.5

  6. Measure the model quality in terms of MSE in train and test samples, (0.5 points) OK || 0.5

  7. Write a short report about the work done. Why did you take these particular features? Can you find a logical explanation for high correlation of some of your features? Are you satisfied with the quality of predictions? etc. (1 point) NOT OK Report is out. || 0

ghost commented 3 years ago

I was a bit lost with the rule I did not know that we have a specific time apart from the date, however I have check all other work on the same date. Maybe you will give a mark for this time, as a lesson learned. Now I am checking other tasks before the specific time. @JackNightsky Please.

JackNightsky commented 3 years ago

I would put 0. After consulting with other activists, given that this is the first homework assignment, we decided to put the mark on which I checked the work - 95. I hope now you check the work on the very first day after they appear for review. @Kaktuscha

Also, your post-deadline review of other work did not affect the other students' scores. Because grades given after the deadline are not counted. So you framed these students.

ghost commented 3 years ago

@JackNightsky Thank you, I will check the remaining cross-checks on time.

dr-leto commented 3 years ago

Issue is verified and score is updated