Python - Sum negative and positive values using GroupBy in Pandas

When working with grouped data in Pandas, you may need to calculate separate sums for positive and negative values within each group. This is useful for analyzing temperature data, financial gains/losses, or any dataset with both positive and negative values.

Creating the DataFrame

First, let's create a DataFrame with temperature data containing both positive and negative values ?

import pandas as pd

# Create DataFrame with temperature data
dataFrame = pd.DataFrame({
    'Place': ['Chicago', 'Denver', 'Atlanta', 'Chicago', 'Dallas', 'Denver', 'Dallas', 'Atlanta'], 
    'Temperature': [-2, 30, -5, 10, 30, -5, 20, -10]
})
print(dataFrame)

     Place  Temperature
0  Chicago           -2
1   Denver           30
2  Atlanta           -5
3  Chicago           10
4   Dallas           30
5   Denver           -5
6   Dallas           20
7  Atlanta          -10

Defining Helper Functions

We need two functions to separate positive and negative values ?

import pandas as pd

def plus(val):
    return val[val > 0].sum()

def minus(val):
    return val[val < 0].sum()

# Test the functions
test_data = pd.Series([-2, 10, -5])
print("Positive sum:", plus(test_data))
print("Negative sum:", minus(test_data))

Positive sum: 10
Negative sum: -7

Grouping and Aggregating

Now we can group by Place and apply both functions to get separate sums ?

import pandas as pd

# Create DataFrame
dataFrame = pd.DataFrame({
    'Place': ['Chicago', 'Denver', 'Atlanta', 'Chicago', 'Dallas', 'Denver', 'Dallas', 'Atlanta'], 
    'Temperature': [-2, 30, -5, 10, 30, -5, 20, -10]
})

# Define functions
def plus(val):
    return val[val > 0].sum()

def minus(val):
    return val[val < 0].sum()

# Group by Place and apply both functions
groupRes = dataFrame.groupby('Place')
result = groupRes['Temperature'].agg([('negTemp', minus), ('posTemp', plus)])
print(result)

         negTemp  posTemp
Place                   
Atlanta      -15        0
Chicago       -2       10
Dallas         0       50
Denver        -5       30

Alternative Approach Using Lambda Functions

You can also use lambda functions directly in the aggregation ?

import pandas as pd

dataFrame = pd.DataFrame({
    'Place': ['Chicago', 'Denver', 'Atlanta', 'Chicago', 'Dallas', 'Denver', 'Dallas', 'Atlanta'], 
    'Temperature': [-2, 30, -5, 10, 30, -5, 20, -10]
})

# Using lambda functions directly
result = dataFrame.groupby('Place')['Temperature'].agg([
    ('negTemp', lambda x: x[x < 0].sum()),
    ('posTemp', lambda x: x[x > 0].sum())
])
print(result)

         negTemp  posTemp
Place                   
Atlanta      -15        0
Chicago       -2       10
Dallas         0       50
Denver        -5       30

Key Points

Conclusion

Use groupby().agg() with custom functions to separate positive and negative value sums. This approach is efficient for analyzing data with mixed positive/negative values across different categories.