求助：使用tidyr的gather函数将数据框转换为长格式

Hey Lisa, no need to stress—this is a super common task in pandas, and we can get it sorted quickly! Let’s walk through this with a concrete example that matches the typical wide-to-long use case, since you mentioned you’re stuck on the conversion.

Example Scenario #

First, let’s assume your original wide-format DataFrame looks like this (since you didn’t share the exact content, I’ll use a relatable example):

user_id  jan_sales  feb_sales  mar_sales
0        1        150        200        180
1        2        300        250        220
2        3        120        160        140

And you generated it with code similar to this:

import pandas as pd

data = {
    'user_id': [1, 2, 3],
    'jan_sales': [150, 300, 120],
    'feb_sales': [200, 250, 160],
    'mar_sales': [180, 220, 140]
}
wide_df = pd.DataFrame(data)

Your target long-format DataFrame is probably structured like this, where each row represents a single user-month sales entry:

user_id    month  sales
0        1      jan    150
1        1      feb    200
2        1      mar    180
3        2      jan    300
4        2      feb    250
5        2      mar    220
6        3      jan    120
7        3      feb    160
8        3      mar    140

The Solution: Use pandas.melt() #

The melt() function is designed exactly for this wide-to-long conversion. Here’s how to apply it to the example above:

long_df = wide_df.melt(
    id_vars=['user_id'],  # Columns to keep as identifiers (not melted)
    var_name='month',     # Name of the new column for the melted headers
    value_name='sales'    # Name of the new column for the values
)

# Optional: Clean up the 'month' column to remove '_sales'
long_df['month'] = long_df['month'].str.replace('_sales', '')

What if Your Use Case is More Complex? #

If your original DataFrame has multiple groups of columns to melt (e.g., both sales and expenses per month), you can use melt() with multiple value_vars or even pandas.wide_to_long() for more structured multi-column conversions. For example:

# If you have both sales and expenses columns
wide_df_exp = pd.DataFrame({
    'user_id': [1,2,3],
    'jan_sales': [150,300,120],
    'jan_expenses': [50,70,40],
    'feb_sales': [200,250,160],
    'feb_expenses': [60,80,50]
})

long_df_exp = pd.wide_to_long(
    wide_df_exp,
    stubnames=['sales', 'expenses'],
    i='user_id',
    j='month',
    sep='_'
).reset_index()

This will give you a clean long format with user_id, month, sales, and expenses columns.

Just adjust the id_vars, var_name, and value_name parameters to match your actual DataFrame structure, and you’ll be set!

内容的提问来源于stack exchange，提问作者Lisa Staal