如何将Pandas DataFrame按单列分组转换为多级索引

Hey there! Let's break down how to achieve exactly what you want with your Pandas DataFrame, step by step.

Step 1: Replicate Your Original DataFrame #

First, let's recreate the DataFrame you provided to work with:

import pandas as pd

data = {
    'group': [1, 1, 1, 2, 2, 3],
    'A': [1, 2, 4, 8, 5, 6],
    'B': [2, 3, 9, 1, 6, 5],
    'C': [3, 6, 9, 2, 4, 7]
}
df = pd.DataFrame(data)

Step 2: Convert to the Desired Display Format #

To get the clean, repeated-index-hidden display you want, set group as the index and enable Pandas' sparse display option:

# Set 'group' as the DataFrame index
df_indexed = df.set_index('group')

# Turn on sparse display to hide duplicate index values in output
pd.set_option('display.multi_sparse', True)

Now when you print df_indexed, it will match your desired format perfectly:

A  B  C
group         
1      1  2  3
       2  3  6
       4  9  9
2      8  1  2
       5  6  4
3      6  5  7

Step 3: Access Sub-DataFrames by Group #

You have a couple of flexible ways to pull up specific group data:

Option 1: Use .loc with Group Labels #

If you want to directly access the sub-DataFrame for group 1, use index-based selection:

group_1_df = df_indexed.loc[1]

This returns all rows for group 1, and you can immediately run operations like mean() on it:

group_1_mean = group_1_df.mean()
# Output: A    2.333333, B    4.666667, C    6.000000

Option 2: Use groupby for Batch or Position-Based Access #

For more control (like accessing groups by position, e.g., 0 for the first group), use groupby:

# Group the original DataFrame by 'group'
grouped = df.groupby('group')

# Get group 1's sub-DataFrame
group_1_df = grouped.get_group(1)

# Access the first group by position (matches group 1 in your data)
first_group_df = list(grouped)[0][1]

Step 4: Run Operations on Sub-DataFrames #

All standard Pandas operations work seamlessly on these sub-DataFrames:

# Calculate mean for group 2
group_2_mean = grouped.get_group(2).mean()

# Calculate sum for group 3
group_3_sum = grouped.get_group(3).sum()

# Compute stats for all groups at once
all_groups_summary = grouped.agg(['mean', 'sum', 'max'])

内容的提问来源于stack exchange，提问作者AstroBen