熊猫找到最近轮廓的行

Question

I have a file full of profiles that look like this:

 profile_id  colA  colB  colC  colD
 1           1     20    50    63
 2           1     20    65    38
 3           8     5     3     4
 4           98    1     878   4
 ...

I have another CSV with results from which I want to find the profile:

col    value    score
colA   1        85
colA   1        856
colA   8        200000
colB   1        2356
colC   878      99999
colD   4        2
...

I want to extract the value for each colX with the best score and find which profile_id this is associated to in the previous file.

What I've done is working:

profiles = pd.read_csv("profiles.csv", sep="\t", index_col=False)
df = pd.read_csv("results.csv", sep="\t", index_col=False)

found_col = set(df["col"])
good_profile = profiles.copy()
for col in profiles.columns:
    if col == "profile_id":
        continue
    elif col not in found_col:
        print(f"{col} not found")
    else:
        value = int(df.loc[df[df["col"] == col]["score"].idxmax()].value)
        good_profile = good_profile[good_profile[col] == value]
 print(good_profile)

This give me the result I want but I am first extracting a subset for the first column, then a subset of this subset for the second one etc...

The cool thing with this is I also get a result when I miss some columns which is great.

I was wondering if there were a way for it to do it better without having to use to create subsets over the previous subset.

Answer 1

Here's my attempt:

# extract the id with max scores
new_df = df2.loc[df2.groupby('col').score.idxmax(), ['col','value']]

# merge
new_df.merge(df1.melt(id_vars='profile_id', var_name='col'),
             on=['col','value'],
             how='left')

Output:

    col  value  profile_id
0  colA      8           3
1  colB      1           4
2  colC    878           4
3  colD      4           3
4  colD      4           4