如何用Pandas将DataFrame指定列转为行？保留time列作为索引

Got it, let's fix this for you! What you're trying to do is reshape your wide dataframe into a long (tidy) format—this is a super common task, and you were already on the right track with melt! The issue was just a tiny parameter mix-up.

Let's Break Down the Solution #

First, let's recap your goal:

• Keep the time column as your identifier (each row in the original data should map to multiple rows in the output, one for each colA-colG).
• Turn the column names colA-colG into values in a new cols column.
• Turn the values under colA-colG into values in a new vals column.

Correct Usage of pd.melt() #

Your earlier melt attempt failed because you used id_vars='time_period'—but your original data's identifier column is named time, not time_period. Here's the fix:

Step 1: Create Example Data (to replicate your scenario)

import pandas as pd

# Your sample data
data = {
    'time': [12, 13],
    'colA': [1, 2],
    'colB': [4, 7],
    'colC': [5, 6],
    'colD': [9, 8],
    'colE': [4, 3],
    'colF': [3, 4],
    'colG': [3, 5]
}
df = pd.DataFrame(data)

Step 2: Reshape with melt

# This will give you exactly the output you want
long_df = df.melt(
    id_vars='time',       # Column to keep as identifier
    var_name='cols',      # Name for the new column holding original column names
    value_name='vals'     # Name for the new column holding the values
)

Step 3: Result

Running the code above will produce this output:

time cols  vals
0    12 colA     1
1    13 colA     2
2    12 colB     4
3    13 colB     7
4    12 colC     5
5    13 colC     6
6    12 colD     9
7    13 colD     8
8    12 colE     4
9    13 colE     3
10   12 colF     3
11   13 colF     4
12   12 colG     3
13   13 colG     5

Why Your Previous Attempts Didn't Work #

• df.transpose(): This just flips the entire table (rows ↔ columns), which turns your time values into row labels and original rows into column names—totally not what you need for this task.
• Pivot tables: pivot() or pivot_table() are used to convert long data to wide data, which is the opposite of your goal, so they aren't the right tool here.
• Your original melt call: The typo in id_vars (using time_period instead of time) was the only issue—fix that and it works perfectly!

Extra: Be Explicit with Columns (Optional) #

If your dataframe has other columns you don't want to include in the reshaping, you can explicitly list the columns to unpivot using value_vars:

long_df = df.melt(
    id_vars='time',
    value_vars=['colA', 'colB', 'colC', 'colD', 'colE', 'colF', 'colG'],
    var_name='cols',
    value_name='vals'
)

This ensures you only reshape the columns you intend to, avoiding any unexpected data in your output.

内容的提问来源于stack exchange，提问作者Emm