python - AttributeError : Can only use . 带有字符串值的 str 访问器，它在 pandas (Python) 中使用 np.object_ dtype

Question

我正在操作一个 JSON 文件，我从中运行此代码以获取以下数据帧:

import pandas as pd

topics = df.set_index('username').popular_board_data.str.extractall(r'name":"([^,]*)')
total = df.set_index('username').popular_board_data.str.extractall(r'totalCount\":([^,}]*)')

data = []
for username in df.username.unique():
for topic in zip(topics[0][username], total[0][username]):
    data.append([username, topic])

df_topic = pd.DataFrame(data, columns='username,topic'.split(','))

    username        topic
0     lukl    (Hardware", 80)
1     lukl    (Marketplace", 31)
2     lukl    (Atari 5200", 27)
3     lukl    (Atari 8-Bit Computers", 9)
4     lukl    (Modern Gaming", 3)

现在我需要将“主题”列中的信息分成两个不同的列:

这是预期的结果:

    username        topic          _topic       _total
0     lukl    (Hardware", 80)      Hardware     80
1     lukl    (Marketplace", 31)   Marketplace  31
2     lukl    (Atari 5200", 27)    Atari 5200   27
3     lukl    (Atari 8", 9)        Atari 8      9
4     lukl    (Modern", 3)         Modern       3

我想用这段代码来做:

df_top = df_topic.copy()
df_top['_topic'] = df_topic['topic'].str.split('(').str[1].str.split('",').str[0]
df_top['_total'] = df_topic['topic'].str.split('",').str[1].str.split(')').str[0]
df_top

但是我收到了这个错误:

AttributeError: Can only use .str accessor with string values, which use np.object_ dtype in pandas

Answer 1

我认为有元组，所以只使用 DataFrame 构造函数:

df_topic[['_topic', '_total']]=pd.DataFrame(df_topic['topic'].values.tolist(), 
                                index=df_topic.index)

更好的解决方案是使用 concat 之前的答案数据和 DataFrame.reset_index :

df = [{"username": "last",
    "popular_board_data": "{\"boards\":[{\"postCount\":\"75\",\"topicCount\":\"5\",\"name\":\"Hardware\",\"url\",\"totalCount\":80},{\"postCount\":\"20\",\"topicCount\":\"11\",\"name\":\"Marketplace\",\"url\",\"totalCount\":31},{\"postCount\":\"26\",\"topicCount\":\"1\",\"name\":\"Atari 5200\",\"url\",\"totalCount\":27},{\"postCount\":\"9\",\"topicCount\":0,\"name\":\"Atari 8\",\"url\"\"totalCount\":9}"
    },
    {"username": "truk",
     "popular_board_data": "{\"boards\":[{\"postCount\":\"351\",\"topicCount\":\"11\",\"name\":\"Atari 2600\",\"url\",\"totalCount\":362},{\"postCount\":\"333\",\"topicCount\":\"22\",\"name\":\"Hardware\",\"url\",\"totalCount\":355},{\"postCount\":\"194\",\"topicCount\":\"8\",\"name\":\"Marketplace\",\"url\",\"totalCount\":202}"
    }]
df = pd.DataFrame(df)

---

#added " for remove it from output
topics = df.set_index('username').popular_board_data.str.extractall(r'name":"([^,]*)"')
total = df.set_index('username').popular_board_data.str.extractall(r'totalCount\":([^,}]*)')

df1 = pd.concat([topics[0], total[0]], axis=1, keys=['_topic', '_total'])
df1 = df1.reset_index(level=1, drop=True).reset_index()
print (df1)
  username       _topic _total
0     last     Hardware     80
1     last  Marketplace     31
2     last   Atari 5200     27
3     last      Atari 8      9
4     truk   Atari 2600    362
5     truk     Hardware    355
6     truk  Marketplace    202