熊猫将浮点数附加到for循环中的列

Question

I have pandas DataFrame where I want to do some calculations using elements in the df, and then append the calculated number into a separate column in the same df.

Here is my code as of now.

def percentfunction(df):
    for i in range(100): 
        if df['month_number'][i] == 10:
            df = df['percent_october'][i].add([df['cellsum'][i]/octobersum])
        elif df['month_number'][i] == 11:
            df = df['percent_november'][i].add([df['cellsum'][i]/novembersum])
        elif df['month_number'][i] == 12:
            df = df['percent_december'][i].add([df['cellsum'][i]/decembersum])

AttributeError: 'numpy.float64' object has no attribute 'add'

I have tried various implementations of this code, but I always get an error message. Either it gets to the last element and then writes the columns containing only the last number calculated, or it adds in rows where it is not supposed to add anything.

Critiques welcome!

EDIT: Tried to edit the code.

    def percentfunction(df):
        for i in range(100): 
            if df['month_number'][i] == 10:
                df['percent_october'][i] = df['cellsum'][i]/octobersum
            elif df['month_number'][i] == 11:
                df['percent_november'][i] = df['cellsum'][i]/novembersum
            elif df['month_number'][i] == 12:
                df['percent_december'][i] = df['cellsum'][i]/decembersum

I get this to run at least, but this fills in values in rows where it shouldn't as well...

EDIT2: Here is a sample of my dataframe

>>> df.head()
      Index          month_number        month_text  \
0     Name1                    10           October     
1     Name1                    11           November    
2     Name1                    12           December    
3     Name2                    10           October     
4     Name2                    11           November    

  2000 Unnamed: 4 2001 Unnamed: 6     2002 Unnamed: 8 2003    ...     \
0  NaN        NaN  NaN        NaN      NaN        NaN  NaN    ...      
1  NaN        NaN  NaN        NaN      NaN        NaN  NaN    ...      
2  NaN        NaN  NaN        NaN      NaN        NaN  NaN    ...      
3  NaN        NaN  NaN        NaN  2898.68       3120  NaN    ...      
4  NaN        NaN  NaN        NaN      NaN        NaN  NaN    ...      

  Unnamed: 28 2013 Unnamed: 30  2014 Unnamed: 32 2015 Unnamed: 34 2016  \
0         NaN  NaN         NaN   NaN         NaN  NaN         NaN  NaN   
1         NaN  NaN         NaN   NaN         NaN  NaN         NaN  NaN   
2         NaN  NaN         NaN   NaN         NaN  NaN         NaN  NaN   
3         NaN  NaN         NaN   NaN         NaN  NaN         NaN  NaN   
4         NaN  NaN         NaN  1.26         127  NaN         NaN  NaN   

  Unnamed: 36   cellsum  
0         NaN      3899  
1         NaN      7922  
2         NaN      2181  
3         NaN      3121  
4         NaN       127

This is my DataFrame, the 'cellsum' is the sum of all the "Unnamed" cells across that row. I have calculated the total month sum by summing all the, for example october cellsums (octobersum) in the DataFrame. I then want to add a new column, with what percentage that cellsum is. I hope you can understand.

Answer 1

You should avoid loops with pandas. You need something like this which you can then manipulate into any format you want:

df["percent_month"] = df.groupby("month_number").apply(lambda x: x/x.sum())

Answer 2

a simple fix would be using df.ix[]

df.ix[i,'percent_october'] = df.ix[i,'cellsum']/octobersum

if you show us what df looks like we might be able to give you a smarter solution than looping over a dataframe which is not recommended