# How to calulate difference between two columns and flag based on condition?

## Question:

I have dataframe

``````  Group  Required  stock
0     A        10      5
1     A        10      8
2     A        10      7
3     B        13      6
4     B        13      5
5     C         8      4
6     C         8      5
7     C         8      8
8     D        16    NaN
``````

Here required for A, B, C, D is `[10,13,8,16]` and my respective stock is mentioned above in table. I need to flag rows what all need to be moved and how many quantity need to be moved

Output should be

``````  Group  Required  stock  to_move flag
0     A        10    5.0      5.0  yes
1     A        10    8.0      5.0  yes
2     A        10    7.0      0.0   no
3     B        13    6.0      6.0  yes
4     B        13    5.0      5.0  yes
5     C         8    4.0      4.0  yes
6     C         8    5.0      4.0  yes
7     C         8    8.0      0.0   no
8     D        16    NaN      NaN   no
``````

You can just assign new columns in pandas:

``````>>> df = pd.DataFrame({'Group': ['A', 'A', 'A', 'B', 'B', 'C', 'C', 'C', 'D']})
>>> df
Group
0     A
1     A
2     A
3     B
4     B
5     C
6     C
7     C
8     D
>>> df['to_move'] = ['Yes']*2+['No']+['Yes']*4+['No']*2
>>> df
Group to_move
0     A     Yes
1     A     Yes
2     A      No
3     B     Yes
4     B     Yes
5     C     Yes
6     C     Yes
7     C      No
8     D      No

``````

Use:

``````#create cumulative sum per groups
s = df.groupby('Group')['stock'].cumsum()
#get difference with Required
diff = df['Required'].rsub(s)
#comapre if difference is less or equal like Stock
m = diff.le(df['stock'])

#subtract stock if diffrence less 0
df['to_move'] = df['stock'].sub(diff.where(diff.gt(0), 0)).where(m, 0)
#create Flag column
df['Flag'] = np.where(m, 'Yes', 'No')

print (df)
Group  Required  stock  to_move Flag
0     A        10    5.0      5.0  Yes
1     A        10    8.0      5.0  Yes
2     A        10    7.0      0.0   No
3     B        13    6.0      6.0  Yes
4     B        13    5.0      5.0  Yes
5     C         8    4.0      4.0  Yes
6     C         8    5.0      4.0  Yes
7     C         8    8.0      0.0   No
8     D        16    NaN      0.0   No
``````

You can use a `groupby.cumsum` with `clip` to compute the cumulated values to move without overflow, then `groupby.diff` to back-calculate the individual values:

``````# compute the cumsum per group
# clip it to not go over the required value
s = df.groupby('Group')['stock'].cumsum().clip(upper=df['Required'].values)

# back calculate the incremental values
df['to_move'] = s.groupby(df['Group']).diff().fillna(s)

# assign the flag if a strictly positive value was moved
df['flag'] = np.where(df['to_move'].gt(0), 'yes', 'no')
``````

Output:

``````  Group  Required  stock  to_move flag
0     A        10    5.0      5.0  yes
1     A        10    8.0      5.0  yes
2     A        10    7.0      0.0   no
3     B        13    6.0      6.0  yes
4     B        13    5.0      5.0  yes
5     C         8    4.0      4.0  yes
6     C         8    5.0      4.0  yes
7     C         8    8.0      0.0   no
8     D        16    NaN      NaN   no
``````
Categories: questions Tags: , , ,
Answers are sorted by their score. The answer accepted by the question owner as the best is marked with
at the top-right corner.