I'm running code that modifies values within a certain threshold in a dataframe. I receive a warning that on the surface does not seem warranted:
SettingWithCopyWarning:
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead
My code already uses the a .loc[row,column] assignment, therefore I do not understand why the warning suggests that.
import pandas as pd
#pd.options.mode.chained_assignment = None #disable warning
#pd.set_option('mode.chained_assignment','warn')#or "warn" or "raise"
u = (df
# Group all forecasts together
.groupby(by="forecast_id", sort=False)
# modify only forecasts groups that have smallest value = 0
.filter(lambda x: x.value.min() == 0, dropna=False)
# transform values according to a function
.value.transform( lambda x: (x+0.005).where(x == 0, x-0.005) )
)
# replace the column in the dataframe with the new values except those unaffected
df.loc[pd.notnull(u), "value"] = u
The other behavior I could not explain is that as I was playing around with the warning options, once I set the Warning to None, even if I reset it back to "warn", there is no warning anymore. Note: my code is used as a function.
Edit
A description of what the code does as well an example, is provided in the link at the top; however, my focus here is understanding why why the warning is suggesting an implementation that is already implemented: Pandas - Calculate New Value Based on Cross Reference with Another Column
