Hi, I have a question, what we have to predict in the te...

대출 상점 총 매출 예측 경진대회

2018.08.17 00:34 6,163 조회

Hi,

I have a question, what we have to predict in the test dataset.

According to my understanding, we need to find the sales of each store for next 100 days after 31st July 2018, which is equivalent to, finding the spending of each card in each store and aggregate the data on the store level for next 100 days.

In the test dataset, Amount column is already present and filled. I am confused now!!

댓글 7개

로그인이 필요합니다

comment

0 / 1000

DACONIO

2022.02.14 16:20

We allow you to check your model with test.csv.

Chiki

2022.02.14 16:20

So if we will find the sales for the next 100 days than, how can they validate our predictions given the last day is 31st July 2018.

1. In test, they have store details till store number 199 and the same with submission, do we need to predict for only those stores?

2. you can find the test and train are overlapping do we need to bind them both for predictions, if so then why there is a test file.

3. Evaluation should be done on real data but real data is not at all available