Pandas：将分组结果转为DataFrame并合并相似索引重复行

Hey there! Let's sort out this duplicate merging task for your DataFrame. I see you want to combine rows that share the same similarity_index—here's how to do it properly:

First, let's recap your original DataFrame for clarity:

import pandas as pd

data = {
    "ad soyad": ["hakan özdemir", "hasan yaman", "naci şenli", "naciye şen", "osman uygur", "elif sözen", "irem derici"],
    "similarity_index": [0, 1, 2, 2, 3, 4, 5]
}
test_df = pd.DataFrame(data)

The code you tried (test_df.set_index("similarity_index").sort_index()) just reorders and indexes your data, but it doesn't actually merge the duplicate rows. To combine rows with matching similarity_index values, we'll use Pandas' groupby method, which is built exactly for this kind of grouping/aggregation task.

Solution 1: Merge names into a single string #

If you want to combine the ad soyad entries into a single comma-separated string for each index, use this:

# Group by similarity_index, join the names, and convert back to a DataFrame
merged_df = test_df.groupby("similarity_index")["ad soyad"].agg(', '.join).reset_index()

This will give you a clean DataFrame where each unique similarity_index has one row with all corresponding names merged:

similarity_index          ad soyad
0                 0      hakan özdemir
1                 1       hasan yaman
2                 2  naci şenli, naciye şen
3                 3       osman uygur
4                 4        elif sözen
5                 5       irem derici

Solution 2: Keep grouped values as lists (for more flexibility) #

If you prefer to keep each original entry as an element in a list (useful if you need to process them later), use agg(list) instead:

merged_df = test_df.groupby("similarity_index").agg(list).reset_index()

The output will look like this:

similarity_index                ad soyad
0                 0      [hakan özdemir]
1                 1       [hasan yaman]
2                 2  [naci şenli, naciye şen]
3                 3       [osman uygur]
4                 4        [elif sözen]
5                 5       [irem derici]

The key here is using groupby to cluster rows by similarity_index, then applying an aggregation function to combine the values in each group. The reset_index() step ensures your result is a proper DataFrame (not a Series with the index as similarity_index).

内容的提问来源于stack exchange，提问作者Hakan Utku Özdemir