Pandas Tricks -Part 1: Get First and Second Max Column for Each Row

import pandas as pd

import numpy as np

# Sample DataFrame

df = pd.DataFrame({

'A': [10, 25, 30],

'B': [20, 15, 10],

'C': [15, 30, 25]

})

print("Original DataFrame:")

print(df)

# First max column name per row

first_max_col = df.idxmax(axis=1)

# Mask the first max to find second max

masked_df = df.mask(df.eq(df.max(axis=1), axis=0))

# Second max column name per row

second_max_col = masked_df.idxmax(axis=1)

# Combine results

result = pd.DataFrame({

'First Max Column': first_max_col,

'Second Max Column': second_max_col

})

print("\nColumn names with first and second max values:")

print(result)

Explanation:

• df.idxmax(axis=1) → gets the column name with the maximum value per row.

• df.max(axis=1) → gets the max value per row.

• df.eq(df.max(axis=1), axis=0) → returns a boolean DataFrame where True indicates the first max in each row.

• df.mask(...) replaces the max values with NaN, so the second max can be found.

• masked_df.idxmax(axis=1) → now gives the column of the second max.

If you want column indexes (integers) instead of names:

first_max_idx = df.values.argmax(axis=1)

# Mask the max values

masked = df.where(df.ne(df.max(axis=1), axis=0))

second_max_idx = masked.values.argmax(axis=1)

print("\nColumn indexes:")

for i, (first, second) in enumerate(zip(first_max_idx, second_max_idx)):

print(f"Row {i}: First max at index {first}, Second max at index {second}")

sorting column names programmatically

import pandas as pd

df = pd.DataFrame({

'SKU': [1, 2],

'Demand': [100, 200],

'RDC': ['A', 'B'],

'Class': ['A', 'B']

})

# Sort columns alphabetically

df = df[sorted(df.columns)]

print(df)