Pandas技术问题：如何展开含不同键字典的DataFrame行？

Hey there! Let's work through this dictionary key mismatch problem you're stuck on. Since you’ve already got the hang of handling cases where the keys in column c are the same, I’ll break down practical approaches for when keys differ—using a super common pandas DataFrame scenario as an example, since that’s where this issue pops up most often.

First, let’s set up a sample input that mirrors your scenario:

import pandas as pd

# Sample input DataFrame
df = pd.DataFrame({
    'a': [1, 2],
    'b': [3, 4],
    'c': [{'x': 5, 'y': 6}, {'y': 7, 'z': 8}]
})

#1: Expand All Unique Keys (Fill Missing Values with NaN) #

This is the go-to approach if you want a wide-format output that includes every unique key from all dictionaries, filling in NaN for rows where a key doesn’t exist.

# Expand the dictionaries in column 'c' into separate columns
expanded_c = pd.json_normalize(df['c'])
# Merge the expanded columns back with the original DataFrame (dropping the original 'c' column)
final_df = pd.concat([df.drop('c', axis=1), expanded_c], axis=1)

Result: #

#2: Convert to Long-Format Key-Value Pairs #

If you prefer a flexible, long-format output that preserves every key-value pair without forcing a unified column structure, this method works great:

# Convert each dictionary into a list of (key, value) tuples
df['c'] = df['c'].apply(lambda x: list(x.items()))
# Explode the list into separate rows, then split into key/value columns
expanded_df = df.explode('c')
expanded_df[['key', 'value']] = pd.DataFrame(expanded_df['c'].tolist(), index=expanded_df.index)
# Clean up by dropping the original tuple column
final_df = expanded_df.drop('c', axis=1)

Result: #

#3: Keep Only Shared Common Keys #

If you only care about keys that exist in all dictionaries in column c, first identify the common keys, then extract just those values:

# Find keys that are present in every non-null dictionary in column 'c'
common_keys = set.intersection(*[set(d.keys()) for d in df['c'].dropna()])
# Extract only the common keys from each dictionary, then expand into columns
expanded_common = df['c'].apply(lambda x: {k: x[k] for k in common_keys}).apply(pd.Series)
# Merge back with original DataFrame
final_df = pd.concat([df.drop('c', axis=1), expanded_common], axis=1)

Result (in our sample, only 'y' is a common key): #

If your use case isn’t pandas (e.g., working with raw lists of dictionaries), the core ideas still apply: either standardize all keys with missing value fills, reshape to key-value pairs, or filter for shared keys. Adjust the code to fit your specific data structure!

内容的提问来源于stack exchange，提问作者shobhu