I am new to pandas and I am trying to catch up on its API design. What I am most interested in is to get a good rule of thumb to predict wheter calling a method on a dataframe will return a new copy of it (that I must assign to a variable) or will modify it inplace. The documentation mentions everywhere that Copy-On-Write will be the future standard, therefore I have enabled it setting pd.options.mode.copy_on_write = True and I am only interested in its behaviour when copy on write is active. Here is an example of the transformations I need to apply to a data set loaded from an Excel sheet. Although the snippet below seems to do what I need, I always have to reassign to the variable df the modified dataframe returned by each method. df = pd.read_excel("my_excel_file.xls", sheet_name="my_sheet", usecols="A:N") # load dataframe from excel sheet df = df.dropna(how='all') # remove empty rows df = df.iloc[:-1,:] # remove last row df.columns.array[0] = "Resource" # change name of the first column df = df.astype({"Resource": int}) # change column type df.columns = df.columns.str.replace('Avg of ', '').str.replace('Others', 'others') # alter column names df = df.set_index("Resource") # use 'Resource' column as index df = df.sort_index(axis=0) # sort df by index value df = df / 100 # divide each entry by 100 df = df.round(4) # round to 4 decimals df = df.reindex(columns=sorted(df)) # order columns in ascending alphabetical order What is the recommended way to carry out the operations in the snippet above? Is it correct to assume that each method that modifies the dataframe is not applied inplace and returns a new dataframe object that I need to assign to a variable? Continue reading...
