[BUG] Cannot use 'Value' as the column name for observation column #857
Conversation
CharlesRendle
changed the title
CharlesRendle
changed the title
src/csvcubed/utils/csvdataset.py
Outdated
| # parameter passed to the melt function to "Not-Value" so that we don't | ||
| # trigger a pandas ValueError. | ||
| value_name = "Value" | ||
| if "Value" in value_cols: |
There was a problem hiding this comment.
Does the pandas ValueError trigger if the column title contains "Value" (e.g. if "Value" in), or only if the title is "Value" exactly (== "Value)? I am assuming it is the latter, since this seems to work and I guess it wouldn't otherwise, but I am just checking because I don't know what the error looks like.
There was a problem hiding this comment.
I don't think you need to set value_name to "Value", you can just go straight to checking if "Value" is in value_cols
There was a problem hiding this comment.
@NickPapONS - Changed to loop through column titles in value_cols instead of only matching "Value" in ..
@SarahJohnsonONS - I think we need to assign a default to value_name because its only if Value is a title in value_cols do we assign a new string to value_name but we pass value_name regardless to melt()
|
Kudos, SonarCloud Quality Gate passed! |
Addresses #856 in which when a column is titled
Valueand we need to use the pandasmelt()function such as when callingtransform_dataset_to_canonical_shape()which causes an error because pandas does not allow a melted dataframe to have a column namedValue.This ticket adds a rather hacky workaround in which we catch any columns called
Valueand give the melt function avalue_nameofNot-Value(which is then used in the melted df). We then rename theNot-Valuecolumn from the melted df and pass this back.Added a unit test to verify this functionality. Compares column names from the original dataframe are a subset of the ones in the melted df.