You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, when trying to use find_representative_periods on a dataframe with periods not starting at one (for example trying to run it for the second month, period 31:60), the function behaves differently based on the combination of distance metric and clustering method used. It will either work as expected or it breaks without a clear error message.
I would suggest to either not allow it and throw and error message in the beginning, making sure that the list of periods starts at 1, or adapt the function to work with period list not starting at 1.
Validation and testing
The error was given when using SqEuclidean() as distance metric and :convex_hull as clustering method.
No error was given when using k_medoids as clustering method with CityBlock(),CosineDist(),SqEuclidean()
I did not debug it any further as I just renamed my periods.
Motivation
It can be relevant for users who want to look at a specific month of the data instead of the whole year and want to find representative periods for that month.
Target audience
No response
Can you help?
No response
The text was updated successfully, but these errors were encountered:
Description
Currently, when trying to use
find_representative_periods
on a dataframe with periods not starting at one (for example trying to run it for the second month, period 31:60), the function behaves differently based on the combination of distance metric and clustering method used. It will either work as expected or it breaks without a clear error message.I would suggest to either not allow it and throw and error message in the beginning, making sure that the list of periods starts at 1, or adapt the function to work with period list not starting at 1.
Validation and testing
The error was given when using
SqEuclidean()
as distance metric and:convex_hull
as clustering method.No error was given when using
k_medoids
as clustering method withCityBlock(),
CosineDist(),
SqEuclidean()
I did not debug it any further as I just renamed my periods.
Motivation
It can be relevant for users who want to look at a specific month of the data instead of the whole year and want to find representative periods for that month.
Target audience
No response
Can you help?
No response
The text was updated successfully, but these errors were encountered: