在python

时间:2018-05-29 14:26:41

标签: python pandas for-loop

假设我有一个数据帧:

df = quandl.get("FRED/DEXBZUS")

输出结果为:

print(df)

    Year    Value
1995-01-02  0.8440
1995-01-03  0.8450
1995-01-04  0.8450
1995-01-05  0.8430
1995-01-06  0.8400
1995-01-09  0.8440
1995-01-10  0.8470
1995-01-11  0.8510

我正在尝试创建一个由变量名称填充的新列:

print(df)

    Year     Value  Variable
1995-01-02  0.8440    df
1995-01-03  0.8450    df
1995-01-04  0.8450    df
1995-01-05  0.8430    df
1995-01-06  0.8400    df
1995-01-09  0.8440    df
1995-01-10  0.8470    df
1995-01-11  0.8510    df

我想在循环过程中使用两个不同的数据帧来执行此操作:

df = quandl.get("FRED/DEXBZUS")
df2 = quandl.get("FRED/DEXBZUS")

data = [df, df2]

for i in data:

dps = []

for i in df:
        d = i.reset_index()
        d = pd.DataFrame(d)
        d['variable'] = [i]

但是我没有在列中获得变量名称。

应该是这样的:

    Year     Value  Variable
1995-01-02  0.8440    df
1995-01-03  0.8450    df
1995-01-04  0.8450    df
1995-01-05  0.8430    df
1995-01-06  0.8400    df
1995-01-09  0.8440    df
1995-01-10  0.8470    df
1995-01-11  0.8510    df


2008-01-02  0.8440    df2
2008-01-03  0.8450    df2
2008-01-04  0.8450    df2
2008-01-05  0.8430    df2
2008-01-06  0.8400    df2
2008-01-09  0.8440    df2
2008-01-10  0.8470    df2
2008-01-11  0.8510    df2

2 个答案:

答案 0 :(得分:0)

不确定这是否是最好的方法,但它确实有效:

In [56]: df_list = []
    ...: for i in locals():
    ...:     try:
    ...:         if type(locals()[i]) == pd.core.frame.DataFrame and not i.startswith('_'):
    ...:             df_list.append(i)            
    ...:     except KeyError:
    ...:         pass  

In [57]: df_list
Out[57]: ['df', 'df2']

In [58]: for d in df_list:
    ...:     locals()[d]['Variable'] = d

In [59]: df
Out[59]: 
         Year  Value Variable
0  1995-01-02  0.844       df
1  1995-01-03  0.845       df
2  1995-01-04  0.845       df
3  1995-01-05  0.843       df
4  1995-01-06  0.840       df
5  1995-01-09  0.844       df
6  1995-01-10  0.847       df
7  1995-01-11  0.851       df

In [60]: df2
Out[60]: 
         Year  Value Variable
0  2008-01-02  0.844      df2
1  2008-01-03  0.845      df2
2  2008-01-04  0.845      df2
3  2008-01-05  0.843      df2
4  2008-01-06  0.840      df2
5  2008-01-09  0.844      df2
6  2008-01-10  0.847      df2
7  2008-01-11  0.851      df2

答案 1 :(得分:0)

要获取变量的名称,我们可以使用this answer中的代码,复制如下:

import inspect


def retrieve_name(var):
        """
        Gets the name of var. Does it from the out most frame inner-wards.
        :param var: variable to get name from.
        :return: string
        """
        for fi in reversed(inspect.stack()):
            names = [var_name for var_name, var_val in fi.frame.f_locals.items() if var_val is var]
            if len(names) > 0:
                return names[0]

这个问题是它在循环说出列表时不会工作,因为你只需要获取局部变量的名称。这与变量名在python中的工作方式有关。变量指向一个对象,即内存中的一个位置,但内存中的位置不会指向后面。这意味着给定一个对象,你无法确定它的名字。 像列表这样的容器也是如此。例如,如果您的列表l包含两个对象a和b l=[a,b],则列表实际上不会保存变量a和b的名称。相反,当您创建列表时,它会记录a和b指向的内存中的位置,即对象而不是名称。

d = 'a'
print(retrieve_name(d))
#'d'
l = [d, d]
print([retrieve_name(element) for element in list ])
#['element', 'element']

话虽这么说,如果你有一个名字和对象的字典,你可以做你要求的:

name_dict = {'df': df, 'df2':df2}
dfs = [frame.assign(Variable=name) for name, frame in name_dict.items()]
combined_df = pd.concat(dfs)

但是,如果您的DataFrame实际上都有不同的数据源,那么可以更轻松地完成所有这些操作。我经常遇到这样的问题:在几个不同的源中存储数据,并且它们的名称例如是文件名。让我们说我有几个.csv文件,我正在从中读取数据,我想将它们全部合并到一个pd.DataFrame中,但希望每行记住它来自哪个文件。

import pandas as pd
#Let's make our two fake csv files a and b:
with open('a.csv', mode='w') as a, open('b.csv', mode='w') as b:
     a.write('col1,col2\n1,1')
     b.write('col1,col2\n2,2')

csv_files = ['a.csv', 'b.csv']
dfs = [pd.read_csv(csv_file).assign(filename=csv_file) for csv_file in csv_files] 
#assign let's you assign the value of a column and returns a DataFrame, so it's 
#great for list comprehensions, in which the df['some_col']='some_var'
#syntax does not work

combined_ab = pd.concat(dfs)
combined_ab
#   col1  col2 filename
#0     1     1    a.csv
#0     2     2    b.csv