Question 9: Choice phrase needs revision?

M4. Linear Models M4. Wrap-up quiz 4
#1

Hi,
The third choice in “Q9” says

“[7, 10]: the linear model is substantially better than the dummy classifier”,

shouldn’t it be “[7,9]” as the index starts from 0 and goes to 9 in 10-folds cross validation?

#2

Here we do not refer to the index but the count of folds.

#3

Let me elaborate my question.
0-3: 4 test scores
4-6: 3 test scores
7-10: 4 test scores

TOTAL: 11 test scores

When I print cv_results(“test_scores”), I get 10 values?
Thus, I think it should be “7-9”? or perhaps “1-3” ? or I am doing something wrong?

#4

You are right. If you run

len(cv_results["test_score"])

you will get a value of 10 that corresponds to the number of folds, as we are using a 10-fold cross-validation (cv=10).

When you compare models A and B fold-to-fold, you could have for instance that the score in the first fold is better for model A

cv_results_A["test_score"][0] > cv_results_B["test_score"][0]

but have that

cv_results_A["test_score"][i] < cv_results_B["test_score"][i]

for i an integer between 1 and 9. This account for the whole 10 folds indexed from 0 to 9. On the other hand, the number of times A is better than B is a number that can run from 0 (A is never better than B) to 10 (A is always better than B), including every value in between, which accounts indeed to 11 possibilities.

#5

Hi Team,

i know you answered the question already, but its still not clear to me …

so we have 10 folds requested through cv=10

so cross_validate would run 10 rounds … right ?

as in python we have indexing from 0, in python “language” we can say round 0 to 9 …

in “cross fold” language we can say round 1 to 10

but options you have provided are [0-10] ( and you used inclusive brackets … i.e. [ and not ( )

my kind question is … when we do cv=10, do we run another 10 rounds? apart from 1 “actual” round … i.e. 11 total rounds, or we run total “10” rounds

thanks team !!!

#6

We do run 10 rounds. The question here does not refer to the count of rounds, but the count of possible answers. Let me list them for extra clarity:

1- A is better than B 1 time;
2- A is better than B 2 times;
3- A is better than B 3 times;
4- A is better than B 4 times;
.
.
.
10- A is better than B 10 times;
11- A is better than B 0 times, i.e. never better than B.