def transform(self, df, **transform_params):
# sanity checks
is_dataframe(df)
print df
return df
def transform(self, df, **transform_params):
# sanity checks
is_dataframe(df)
# if the lag window leads to no successful joins on date, this just returns an empty dataframe
# should that be a test?
return pd.concat([df.shift(l).add_suffix("_L" + str(l)) for l in range(0, self.lag + 1)], axis=1).dropna()
def test_is_dataframe(self):
self.assertRaises(NotDataFrame, is_dataframe, self.not_a_df)
# df1 is a dataframe, so is_dataframe(self.df1) should not raise an exception
# if it does, then this is a problem
try:
is_dataframe(self.df1)
except NotDataFrame:
self.fail("is_dataframe() raised NotDataFrame unexpectedly!")
def transform(self, df, **transform_params):
# sanity checks
is_dataframe(df)
is_column_subset( self.col_labels, df.columns.tolist() )
# extract columns by label
return df[self.col_labels]
def transform(self, df, **transform_params):
# sanity checks
is_dataframe(df)
is_correct_length(df,2)
# Should we be checking for divide by zero?
# It's unlikely in practice with real data, but what if something funky gets passed to this function?
return pd.DataFrame((df.iloc[:,1].sub(df.iloc[:,0])).div(df.iloc[:,0]))
