I have a question regarding the time range of the predicted data.
It is written in the "Task description": " [...] They should be followed by 168 numeric columns containing predictions – mean values of the corresponding characteristics for the next 168 hours (one week starting at 2020-02-20 12:00:00) after the end of the training data."
However, the last timestamp for data points is 2020-02-20T10:00:00Z, which, according to the introduction, contains values from the time range 2020-02-20 10:00-11:00. The next timestamp value is thus 11:00, not 12:00. So, should we:
- Predict " for the next 168 hours ( one week starting at 2020-02-20 12:00:00) after the end of the training data", i.e.
2020-02-20 12:00, 2020-02-20 13:00, ...
or - Predict " for the next 168 hours (one week starting at 2020-02-20 12:00:00) after the end of the training data" i.e.
2020-02-20 11:00, 2020-02-20 12:00, ...
(Edit for the question, as the system did not show the formatting - it should be as below)
- Predict "for one week starting at 2020-02-20 12:00:00", i.e.
2020-02-20 12:00, 2020-02-20 13:00, ...
or - Predict " for the next 168 hours after the end of the training data" i.e.
2020-02-20 11:00, 2020-02-20 12:00, ...
Hello,
thanks for noting this issue!
There is indeed an inconsistency between the timestamps of training and test data resulting from different conventions of time notation. The times in the training data are GMT while in the test data, they are CET (thus the one hour difference).
Since it is misleading, to avoid further confusion we changed the format of timestamps in the training set to CET.
Andrzej Janusz
