Pandas Group Data and Aggregate Variables and Find Duplicates-CodePudding

I have the following data:

plate,part,posX,posY,rotation
1,FSHN01-R-58.stl,14.5,9.5,180
1,FSHN01-R-58.stl,14.5,9.5,180
1,FSHN01-E-2.stl,44.5,6.5,270
1,FSHN01-N-3.stl,88,7,0
2,FSHN01-N-7.stl,70.5,70.5,90
2,FSHN01-N-1.stl,128.5,64.5,180
2,FSHN01-N-1.stl,113.5,69.5,90
7,FSHN01-R-58.stl,14.5,9.5,180
7,FSHN01-R-58.stl,14.5,9.5,180
7,FSHN01-E-2.stl,44.5,6.5,270
7,FSHN01-N-3.stl,88,7,0

I want to group plates by part and find same plates. For example on this data 1,7 have same parts. posX, posY, rotation its not important for me and i need to find part counts of each plates.

For this, i wrote this code:

import pandas as pd

df = pd.read_csv("public/33/stls/plates.csv")


result = df.groupby(['plate', 'part']).size()

print(result)

and i got.

plate  part           
1      FSHN01-E-2.stl     1
       FSHN01-N-3.stl     1
       FSHN01-R-58.stl    2
2      FSHN01-N-1.stl     2
       FSHN01-N-7.stl     1
7      FSHN01-E-2.stl     1
       FSHN01-N-3.stl     1
       FSHN01-R-58.stl    2

so, 1. and 7. plates are same, how can i drop 7. plate from table and increase 1. plate plate count variable.

i need this result;

plate  part           part count  plate count      
1      FSHN01-E-2.stl     1           2
       FSHN01-N-3.stl     1           2
       FSHN01-R-58.stl    2           2
2      FSHN01-N-1.stl     2           1
       FSHN01-N-7.stl     1           1

CodePudding user response：

IIUC, you need two groupby, one with the default plate/part, and one in which you have substituted the plate values 7 by 1:

synonyms = {7: 1}

(df.groupby(['plate', 'part']).size().to_frame(name='part_count')
   .join(df.assign(plate2=df['plate'].replace(synonyms))
           .groupby(['plate2', 'part'])['plate'].agg(plate_count='nunique')
           .rename_axis(['plate', 'part']),
         how='inner'
        )
   .reset_index()
)

or in a more linear form:

synonyms = {7: 1}

df2 = df.groupby(['plate', 'part']).size().to_frame(name='part_count')

df2['plate_count'] = (df.assign(plate2=df['plate'].replace(synonyms))
                        .groupby(['plate2', 'part'])['plate'].nunique()
                     )

df2 = df2.dropna().reset_index()

output:

   plate             part  part_count  plate_count
0      1   FSHN01-E-2.stl           1            2
1      1   FSHN01-N-3.stl           1            2
2      1  FSHN01-R-58.stl           2            2
3      2   FSHN01-N-1.stl           2            1
4      2   FSHN01-N-7.stl           1            1

NB. Note that you voluntarily lose the information on the part counts of plate 7 here. If you want to keep it, map 7 to 1 as shown above and perform a single GroupBy.agg

CodePudding user response：

import pandas as pd

df = pd.read_csv("plates.csv")
df = df.groupby(['plate', 'part']).size().reset_index(name='count')


df_dict = df.groupby('plate')[['part', 'count']].apply(lambda x: x.to_dict('records')).to_dict()


result, plate_count = {}, {}
for key, value in df_dict.items():
    if value not in result.values():
        result[key] = value
        plate_count[key] = 1
    else:
        plate_count[list(result.keys())[list(result.values()).index(value)]]  = 1

print(result)
print(plate_count)

and the output I got:

{1: [{'part': 'FSHN01-E-2.stl', 'count': 1}, {'part': 'FSHN01-N-3.stl', 'count': 1}, {'part': 'FSHN01-R-58.stl', 'count': 2}], 2: [{'part': 'FSHN01-N-1.stl', 'count': 2}, {'part': 'FSHN01-N-7.stl', 'count': 1}]}
{1: 2, 2: 1}

I found a way for fix this issue but i could not fix my problem with pandas and i used dict. I wonder about the better way, anyone can find it ?