correlation shows label 'index' is not unique error

[reza1615]

my index has unique data but it shows the below error.
In my opinion, if the index has duplicated still it should force to show the unique

[aschonfeld]

@reza1615 can you give me some more information about the structure of your data?

• is this timeseries data?
• or is this a correlation of two columns as a whole?

[reza1615]

1-it is tubular data for binary classification like housing data
2-yes it is correlation of two columns as whole and it shows for all columns

[reza1615]

I figured out when I delete the index it will resolved.
In my opinion on fly you can run the below snippet before running the correlation code.
df.index = range(len(df))

[aschonfeld]

So the issue was that when I was calculating the scatter I created a temporary column called index and so if there was an additional column named index being used in the correlation then that would cause our sorting issue with non-unique column names.

I have since updated the code to use _scatter_index (which still may cause a non-unique issue if there is a column of that name) which should be far less commonly used.

[aschonfeld]

fixed in v1.48.0