Closed paulf81 closed 1 month ago
The previous pull request #223 assumed the observed keyword would not have an effect on the computation, and so went for the incoming pandas default value of True, based on my read of: https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.groupby.html
However, this really did change the behavior, which I now understand is because of the preceding cut line:
ws_bin_cuts_subset = [ pd.cut(df[f"ws_{ti:03d}"], bins=self._pw_curve_ws_bins) for ti in turbine_subset ]
ws is a categorical in this context. Therefore reverting back to the current default value made explicit by setting observed=False
observed=False
The previous pull request #223 assumed the observed keyword would not have an effect on the computation, and so went for the incoming pandas default value of True, based on my read of: https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.groupby.html
However, this really did change the behavior, which I now understand is because of the preceding cut line:
ws is a categorical in this context. Therefore reverting back to the current default value made explicit by setting
observed=False