在python中使用pandas加載csv文件

Question

這是我的示例數據：

2017-11-27T00:29:37.698-06:00,,"42,00,00,00,3E,51,1B,D7,42,1C,00,00,40"
2017-11-27T00:29:37.698-06:00,,"42,00,00,00,3E,51,1B,D7,42,1C,00,00,40"
2017-11-27T00:29:37.698-06:00,,"42,00,00,00,3E,51,1B,D7,42,1C,00,00,40"

我嘗試使用pandas加載數據：

data = pd.read_csv("sample.csv",header = None)

我的輸出是：

                0                 1           2
0  2017-11-27T00:29:37.698-06:00 NaN  42,00,00,00,3E,51,1B,D7,42,1C,00,00,40
1  2017-11-27T00:29:37.698-06:00 NaN  42,00,00,00,3E,51,1B,D7,42,1C,00,00,40
2  2017-11-27T00:29:37.698-06:00 NaN  42,00,00,00,3E,51,1B,D7,42,1C,00,00,40

我想將第二列中的每個數據與第一列分開作為時間戳。

我的預期輸出是：

    0                             1  2  3  4....
0  2017-11-27T00:29:37.698-06:00  42 00 00 00
1  2017-11-27T00:29:37.698-06:00  42 00 00 00
2  2017-11-27T00:29:37.698-06:00  42 00 00 00

Answer 1

使用正則表達式傳遞sep參數。 然后，對數據進行一些清理。

df = pd.read_csv(
      'file.csv', 
      sep='"*,',           # separator
      header=None,         # no headers
      engine='python',     # allows a regex with multiple characters
      index_col=[0]        # specify timestamp as the index
)   

df.iloc[:, 1] = df.iloc[:, 1].str.strip('"').astype(int)
df.iloc[:, -1] = df.iloc[:, -1].str.strip('"').astype(int)

df

                               1   2   3   4   5   6   7   8   9   10  11  12  \
0                                                                               
2017-11-27T00:29:37.698-06:00 NaN  42   0   0   0  3E  51  1B  D7  42  1C   0   
2017-11-27T00:29:37.698-06:00 NaN  42   0   0   0  3E  51  1B  D7  42  1C   0   
2017-11-27T00:29:37.698-06:00 NaN  42   0   0   0  3E  51  1B  D7  42  1C   0   

                               13  14  
0                                      
2017-11-27T00:29:37.698-06:00   0  40  
2017-11-27T00:29:37.698-06:00   0  40  
2017-11-27T00:29:37.698-06:00   0  40  

要使用NaNs刪除列，請使用dropna -

df.dropna(how='all', axis=1, inplace=True)

Answer 2

首先添加參數parse_dates=[0]以解析第一列到datetime。

然后join原始split列2並刪除第1列和第2列，最后使用add 1 rename所有列：

df = pd.read_csv("sample.csv",header = None, parse_dates=[0])

df = (df.drop([1,2], axis=1)
        .join(df[2].str.split(',', expand=True)
        .rename(columns = lambda x: x+1))   
      )  
print (df)
                       0   1   2   3   4   5   6   7   8   9   10  11  12  13
0 2017-11-27 06:29:37.698  42  00  00  00  3E  51  1B  D7  42  1C  00  00  40
1 2017-11-27 06:29:37.698  42  00  00  00  3E  51  1B  D7  42  1C  00  00  40
2 2017-11-27 06:29:37.698  42  00  00  00  3E  51  1B  D7  42  1C  00  00  40

詳情

print (df[2].str.split(',', expand=True))
   0   1   2   3   4   5   6   7   8   9   10  11  12
0  42  00  00  00  3E  51  1B  D7  42  1C  00  00  40
1  42  00  00  00  3E  51  1B  D7  42  1C  00  00  40
2  42  00  00  00  3E  51  1B  D7  42  1C  00  00  40

Answer 3

如果需要，您可以執行自己的csv解析器，如：

碼：

def read_my_csv(filename):
    with open(filename, 'rU') as f:

        # build csv reader
        reader = csv.reader(f)

        # for each row, check for footer
        for row in reader:
            yield [row[0]] + row[2].split(',')

測試代碼：

import csv
import pandas as pd

df = pd.DataFrame(read_my_csv('csvfile.csv'))
print(df)

結果：

                                  0   1   2   3   4   5   6   7   8   9   10  \
0      2017-11-27T00:29:37.698-06:00  42  00  00  00  3E  51  1B  D7  42  1C   
1      2017-11-27T00:29:37.698-06:00  42  00  00  00  3E  51  1B  D7  42  1C   
2      2017-11-27T00:29:37.698-06:00  42  00  00  00  3E  51  1B  D7  42  1C   

   11  12  13  
0  00  00  40  
1  00  00  40  
2  00  00  40