Aggregated updates for the next release (0.7.0.3)

• Improvement to RooProcessor and RDataFrame utils
• Add more ROOT-related diagnostic methods
• Fix error in formula string in newer ROOT version
• Fix warning in analysis data preprocessing
• Improvement legend handling
• Fix p-value and significance calculation for any ndof and mu hypothesis in likelihood fits
• Improvement to VariableDistributionPlot: added support for drawing errors on stacked plot, better merging of styles and options, allow hiding portion of data

quickstats/analysis/data_preprocessing.py

@@ -16,9 +16,9 @@ def fix_negative_weights(df, mode:Union[int, NegativeWeightMode, str]=0,
         return None
     mask = df[weight_col] < 0
     if mode == NegativeWeightMode.SETZERO:
-        df[weight_col][mask] = 0
+        df.loc[mask, weight_col] = 0
     elif mode == NegativeWeightMode.SETABS:
-        df[weight_col][mask] = abs(df[weight_col][mask])
+        df.loc[mask, weight_col] = abs(df[weight_col][mask])
         
 def shuffle_arrays(*arrays, random_state:Optional[int]=None):
     if random_state < 0: