如何以一种快速和 Pythonic 的方式在 python pandas 中适应和使用 Excel 的 Sumif function？

Question

I'm working on a project that is trying to port over functions and operations from Excel to python in pandas.

I have a lot of SUMIF functions I'm trying to replicate in the data. Trouble is that I don't think pandas has a specifically analagous function. I might have an excel expression like:

=(SUMIFS('Sheetname':BI$146,BI$282:'Sheetname',$B$146:$B$282,$I1836))

Where the first argument is the region that needs to be summed up. The second region is the range where we're checking for the matching criteria, and the last argument is the specific value we're looking for.

What I'm doing right now is running a nested loop that iterates over all the rows and columns and checks the first iteration finds the matching rows while the inner loop finds the matching columns. The values are then summed and inputted into the pandas function.

Something like:

table_dict_temp is the table I'm populating the values in table_temp is the table to be referenced

for i in range(len(table_dict_temp)):
    cog_loss = table_temp.loc[table_temp[COLUMN OF COMPARISON]==table_dict_temp[COLUMN OF COMPARISON][i]]
    for j in range(10, len(table_dict_temp.columns)):
        cog_loss_temp = cog_loss[table_dict_temp.columns[j]].sum()
        table_dict_temp.iloc[i,j]=cog_loss_temp

The problem I'm running into is that this seems to be a non pythonic way to do this and it takes a lot of time as well. Any advice on how I can write the functioin to be faster would be greatly appreciated!

Answer 1

The Excel example data:

https://support.microsoft.com/en-us/office/sumifs-function-c9e748f5-7ea7-455d-9406-611cebce642b

Quantity Sold   Product Salesperson
5   Apples  Tom
4   Apples  Sarah
15  Artichokes  Tom
3   Artichokes  Sarah
22  Bananas Tom
12  Bananas Sarah
10  Carrots Tom
33  Carrots Sarah
    
Description

=SUMIFS(A2:A9, B2:B9, "=A*", C2:C9, "Tom")

Adds the number of products that begin with A and were sold by Tom. 
It uses the wildcard character * in Criteria1, "=A*" to look for matching product names in Criteria_range1 B2:B9, 
and looks for the name "Tom" in Criteria_range2 C2:C9. 
It then adds the numbers in Sum_range A2:A9 that meet both conditions. 
The result is 20.

=SUMIFS(A2:A9, B2:B9, "<>Bananas", C2:C9, "Tom")

Adds the number of products that aren’t bananas and are sold by Tom. 
It excludes bananas by using <> in the Criteria1, "<>Bananas", 
and looks for the name "Tom" in Criteria_range2 C2:C9. 
It then adds the numbers in Sum_range A2:A9 that meet both conditions. 
The result is 30.

the pythonic solution:

import io
import pandas as pd

data_str = '''
Quantity Sold   Product Salesperson
5   Apples  Tom
4   Apples  Sarah
15  Artichokes  Tom
3   Artichokes  Sarah
22  Bananas Tom
12  Bananas Sarah
10  Carrots Tom
33  Carrots Sarah
'''.strip()

df = pd.read_csv(io.StringIO(data_str), sep='\t')

# =SUMIFS(A2:A9, B2:B9, "=A*", C2:C9, "Tom")
cond = True
cond &= df['Product'].str.startswith('A')
cond &= df['Salesperson'] == 'Tom'
df.loc[cond, 'Quantity Sold'].sum()

# =SUMIFS(A2:A9, B2:B9, "<>Bananas", C2:C9, "Tom")
cond = True
cond &= df['Product'] != 'Bananas'
cond &= df['Salesperson'] == 'Tom'
df.loc[cond, 'Quantity Sold'].sum()