python-3.x 查找pandas datafarme中列之间的关系

6rvt4ljy 于 2023-06-25 发布在 Python

关注(0)|答案(2)|浏览(89)

假设我有一个这样的 Dataframe ：

RING    CLLI    RR          root            CIRCUIT
N100    M200    200.1       OC1             Circuit1
N100    M200    200.1       OC1             Circuit2
N100    M201    200.2       OC1             Circuit3
N100    M202    200.3       OC1             Circuit1
N101    M300    300.1       OC2             Circuit1
N101    M304    301.8       OC2             Circuit11
N101    M147    500.5       OC2             Circuit10
N102    M874    568.7       OC4             Circuit11
N102    M874    568.7       OC4             Circuit114
N102    M874    568.7       OC4             Circuit113
N102    M874    568.7       OC4             Circuit112
N104    M643    414.1       OC8             Circuit2
N104    M643    414.1       OC8             Circuit234
N104    M643    414.1       OC8             Circuit11

我想检查例如，如果列电路在其他行重复.如果它重复，并且RING是不同的，我想添加另一列，并告诉什么是该电路的CLLI。最后，最终的dataframe看起来像这样：

RING    CLLI    RR          root            CIRCUIT           NeigbourCLLI
N100    M200    200.1       OC1             Circuit1            M300
N100    M200    200.1       OC1             Circuit2            M643
N100    M201    200.2       OC1             Circuit3            NaN
N100    M202    200.3       OC1             Circuit1            NaN
N101    M300    300.1       OC2             Circuit1            M200
N101    M304    301.8       OC2             Circuit11           M874, M643          
N101    M147    500.5       OC2             Circuit10           NaN
N102    M874    568.7       OC4             Circuit11           M304, M643
N102    M874    568.7       OC4             Circuit114          NaN         
N102    M874    568.7       OC4             Circuit113          NaN
N102    M874    568.7       OC4             Circuit112          NaN
N104    M643    414.1       OC8             Circuit2            M200
N104    M643    414.1       OC8             Circuit234          NaN
N104    M643    414.1       OC8             Circuit11           M874, M304

我试过这段代码，但它不是那么好，它也迭代了很多，我的 Dataframe 是巨大的，它需要很多时间：

ring_clli_dict = {}
for index, row in df.iterrows():
    circuit = row['CIRCUIT']
    ring = row['RING']
    clli = row['CLLI']
    if circuit in ring_clli_dict:
        if ring != ring_clli_dict[circuit]['RING']:
            ring_clli_dict[circuit]['NeighbourCLLI'].append(clli)
    else:
        ring_clli_dict[circuit] = {'RING': ring, 'NeighbourCLLI': [clli]}

for index, row in df.iterrows():
    circuit = row['CIRCUIT']
    if circuit in ring_clli_dict:
        df.at[index, 'NeighbourCLLI'] = ', '.join(ring_clli_dict[circuit]['NeighbourCLLI'])

有没有更好的办法解决这个问题？

UPD：

好吧，这个很好用，但是正如我提到的，我的DF是50万行，使用这样的代码，需要很长时间才能完成。

grouped = df.groupby('CIRCUIT').apply(lambda x: x['RING'].nunique() > 1)

def find_neighboring_cllis(row):
    circuit, ring = row['CIRCUIT'], row['RING']
    if grouped[circuit]:
        neighbors = df[(df['CIRCUIT'] == circuit) & (df['RING'] != ring)]['CLLI'].unique()
        if neighbors.size > 0:
            return ', '.join(neighbors)
    return np.nan

df['NeigbourCLLI'] = df.apply(find_neighboring_cllis, axis=1)

有没有可能让它更快，并使用更“Pandas”的解决方案，而不是一行一行？

python-3.x

来源：https://stackoverflow.com/questions/76547403/find-relations-between-columns-in-pandas-datafarme

2条答案

按热度按时间

gtlvzcf81#

试试这个：

# Using top 5 rows only:
data={"RING":["N100","N100","N100","N101"],
"CLLI":["M200","M200","M201","M300"],
"RR":["200.1","200.1","200.2","300.1"],
"root":["OC1","OC1","OC1","OC2"],
"CIRCUIT":["Circuit1","Circuit2","Circuit3","Circuit1"]}
df=pd.DataFrame(data)

# Creating a dictionary
to_map=df.groupby("CIRCUIT")["CLLI"].agg(set).to_dict()
df["NeigbourCLLI"]=df["CIRCUIT"].map(to_map)

df['NeigbourCLLI']=df.apply(lambda x:''.join([i for i in x.NeigbourCLLI if i not in x.CLLI]),axis=1)
df['NeigbourCLLI'].replace('',np.nan,inplace=True)

print(df)
    RING    CLLI    RR  root    CIRCUIT NeigbourCLLI
0   N100    M200    200.1   OC1 Circuit1    M300
1   N100    M200    200.1   OC1 Circuit2    NaN
2   N100    M201    200.2   OC1 Circuit3    NaN
3   N101    M300    300.1   OC2 Circuit1    M200

赞(0）回复(0）举报 2023-06-25

4nkexdtk2#

试试这个：

s = df.drop_duplicates(['RING','CIRCUIT']).groupby('CIRCUIT')['CLLI'].transform(lambda x: [set(x)]*x.size)
df.assign(NeigbourCLLI = (s - df['CLLI'].map(lambda x: set([x]))).str.join(',').where(lambda x: x.ne('')))

输出：

RING  CLLI     RR root     CIRCUIT NeigbourCLLI
0   N100  M200  200.1  OC1    Circuit1         M300
1   N100  M200  200.1  OC1    Circuit2         M643
2   N100  M201  200.2  OC1    Circuit3             
3   N101  M300  300.1  OC2    Circuit1         M200
4   N101  M304  301.8  OC2   Circuit11    M874,M643
5   N101  M147  500.5  OC2   Circuit10             
6   N102  M874  568.7  OC4   Circuit11    M643,M304
7   N102  M874  568.7  OC4  Circuit114             
8   N102  M874  568.7  OC4  Circuit113             
9   N102  M874  568.7  OC4  Circuit112             
10  N104  M643  414.1  OC8    Circuit2         M200
11  N104  M643  414.1  OC8  Circuit234             
12  N104  M643  414.1  OC8   Circuit11    M874,M304

赞(0）回复(0）举报 2023-06-25

我来回答

python-3.x 查找pandas datafarme中列之间的关系

2条答案

相关问题

热门标签

最新问答