使用两个分隔符将 CSV 导入 pandas

Import CSV to pandas with two delimiters

我有一个带有两个分隔符 (;) 和 (,) 的 CSV,它看起来像这样:

vin;vorgangid;eventkm;D_8_lamsoni_w_time;D_8_lamsoni_w_value
V345578;295234545;13;-1000.0,-980.0;7.9921875,11.984375
V346670;329781064;13;-960.0,-940.0;7.9921875,11.984375

我想将其导入 pandas 数据框,其中 (;) 作为列分隔符,(,) 作为 [=17] 的分隔符=] 或 array 使用 float 作为数据类型。到目前为止,我正在使用这种方法,但我相信还有更简单的方法。

aa=0;
csv_import=pd.read_csv(folder+FileName, ';')
for col in csv_import.columns:
aa=aa+1
if type(csv_import[col][0])== str and aa>3:
    # string to list of strings
    csv_import[col]=csv_import[col].apply(lambda x:x.split(','))
    # make the list of stings into a list of floats
    csv_import[col]=csv_import[col].apply(lambda x: [float(y) for y in x])

首先使用 ; 作为分隔符读取 CSV:

df = pd.read_csv(filename, sep=';')

更新:

In [67]: num_cols = df.columns.difference(['vin','vorgangid','eventkm'])

In [68]: num_cols
Out[68]: Index(['D_8_lamsoni_w_time', 'D_8_lamsoni_w_value'], dtype='object')

In [69]: df[num_cols] = (df[num_cols].apply(lambda x: x.str.split(',', expand=True)
   ....:                                               .stack()
   ....:                                               .astype(float)
   ....:                                               .unstack()
   ....:                                               .values.tolist())
   ....:                )

In [70]: df
Out[70]:
       vin  vorgangid  eventkm D_8_lamsoni_w_time     D_8_lamsoni_w_value
0  V345578  295234545       13  [-1000.0, -980.0]  [7.9921875, 11.984375]
1  V346670  329781064       13   [-960.0, -940.0]  [7.9921875, 11.984375]

In [71]: type(df.loc[0, 'D_8_lamsoni_w_value'][0])
Out[71]: float

旧答案:

现在我们可以将数字拆分为 "number" 列中的列表:

In [20]: df[['D_8_lamsoni_w_time',  'D_8_lamsoni_w_value']] = \
    df[['D_8_lamsoni_w_time',  'D_8_lamsoni_w_value']].apply(lambda x: x.str.split(','))
In [21]: df
Out[21]:
       vin  vorgangid  eventkm D_8_lamsoni_w_time     D_8_lamsoni_w_value
0  V345578  295234545       13  [-1000.0, -980.0]  [7.9921875, 11.984375]
1  V346670  329781064       13   [-960.0, -940.0]  [7.9921875, 11.984375]

您可以在 read_csv 中使用参数 converters 并定义用于拆分的自定义函数:

def f(x):
    return [float(i) for i in x.split(',')]

#after testing replace io.StringIO(temp) to filename
df = pd.read_csv(io.StringIO(temp), 
                 sep=";", 
                 converters={'D_8_lamsoni_w_time':f, 'D_8_lamsoni_w_value':f})
print (df)
       vin  vorgangid  eventkm D_8_lamsoni_w_time     D_8_lamsoni_w_value
0  V345578  295234545       13  [-1000.0, -980.0]  [7.9921875, 11.984375]
1  V346670  329781064       13   [-960.0, -940.0]  [7.9921875, 11.984375]

4.5. 列中使用 NaN 的另一个解决方案:

您可以使用带有分隔符 ;read_csv,然后将 str.split 应用于 4.5. iloc 选择的列并将 list 中的每个值转换为 float:

import pandas as pd
import numpy as np
import io

temp=u"""vin;vorgangid;eventkm;D_8_lamsoni_w_time;D_8_lamsoni_w_value
V345578;295234545;13;-1000.0,-980.0;7.9921875,11.984375
V346670;329781064;13;-960.0,-940.0;7.9921875,11.984375"""
#after testing replace io.StringIO(temp) to filename
df = pd.read_csv(io.StringIO(temp), sep=";")

print (df)
       vin  vorgangid  eventkm D_8_lamsoni_w_time  D_8_lamsoni_w_value
0  V345578  295234545       13     -1000.0,-980.0  7.9921875,11.984375
1  V346670  329781064       13      -960.0,-940.0  7.9921875,11.984375

#split 4.th and 5th column and convert to numpy array
df.iloc[:,3] = df.iloc[:,3].str.split(',').apply(lambda x: [float(i) for i in x])
df.iloc[:,4] = df.iloc[:,4].str.split(',').apply(lambda x: [float(i) for i in x])
print (df)
       vin  vorgangid  eventkm D_8_lamsoni_w_time     D_8_lamsoni_w_value
0  V345578  295234545       13  [-1000.0, -980.0]  [7.9921875, 11.984375]
1  V346670  329781064       13   [-960.0, -940.0]  [7.9921875, 11.984375]

如果需要 numpy arrays 而不是 lists:

#split 4.th and 5th column and convert to numpy array
df.iloc[:,3] = df.iloc[:,3].str.split(',').apply(lambda x: np.array([float(i) for i in x]))
df.iloc[:,4] = df.iloc[:,4].str.split(',').apply(lambda x: np.array([float(i) for i in x]))
print (df)
       vin  vorgangid  eventkm D_8_lamsoni_w_time     D_8_lamsoni_w_value
0  V345578  295234545       13  [-1000.0, -980.0]  [7.9921875, 11.984375]
1  V346670  329781064       13   [-960.0, -940.0]  [7.9921875, 11.984375]

print (type(df.iloc[0,3]))
<class 'numpy.ndarray'>

我尝试改进您的解决方案:

a=0;
csv_import=pd.read_csv(folder+FileName, ';')
for col in csv_import.columns:
    a += 1
    if type(csv_import.ix[0, col])== str and a>3:
        # string to list of strings
        csv_import[col]=csv_import[col].apply(lambda x: [float(y) for y in x.split(',')])

除了这里的其他更 pandas 具体的好答案之外,应该注意 Python 本身在字符串处理方面非常强大。您可以将 ';' 替换为 ',' 的结果放在 StringIO 对象中,并从那里正常工作:

In [8]: import pandas as pd

In [9]: from cStringIO import StringIO

In [10]: pd.read_csv(StringIO(''.join(l.replace(';', ',') for l in open('stuff.csv'))))
Out[10]: 
                   vin  vorgangid  eventkm  D_8_lamsoni_w_time  \
V345578 295234545   13    -1000.0   -980.0            7.992188   
V346670 329781064   13     -960.0   -940.0            7.992188   

                   D_8_lamsoni_w_value  
V345578 295234545            11.984375  
V346670 329781064            11.984375