pandas 熊猫数据帧到键值对
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/43185659/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
pandas dataframe to key value pair
提问by Sun
What is the best way to convert following pandas dataframe to a key value pair
将以下Pandas数据帧转换为键值对的最佳方法是什么
Before :
前 ：
datetime name qty price
2017-11-01 10:20 apple 5 1
2017-11-01 11:20 pear 2 1.5
2017-11-01 13:20 banana 10 5
After :
后 ：
2017-11-01 10:20 name=apple qty=5 price=1
2017-11-01 11:20 name=pear qty=2 price=1.5
2017-11-01 13:20 name=banana qty=10 price=5
note that i don't want the datetime key in my output.
请注意，我不希望在我的输出中使用 datetime 键。
回答by jezrael
It seems you need to_dict:
看来你需要to_dict：
d = df.drop('datetime', axis=1).to_dict(orient='records')
print (d)
[{'qty': 5, 'price': 1.0, 'name': 'apple'},
{'qty': 2, 'price': 1.5, 'name': 'pear'},
{'qty': 10, 'price': 5.0, 'name': 'banana'}]
but if need not key datetime:
但如果不需要关键datetime：
d = df.set_index('datetime').to_dict(orient='index')
print (d)
{'2017-11-01 13:20': {'qty': 10, 'price': 5.0, 'name': 'banana'},
'2017-11-01 10:20': {'qty': 5, 'price': 1.0, 'name': 'apple'},
'2017-11-01 11:20': {'qty': 2, 'price': 1.5, 'name': 'pear'}}
If order is important:
如果顺序很重要：
tuples = [tup for tup in df.set_index('datetime').itertuples()]
print (tuples)
[Pandas(Index='2017-11-01 10:20', name='apple', qty=5, price=1.0),
Pandas(Index='2017-11-01 11:20', name='pear', qty=2, price=1.5),
Pandas(Index='2017-11-01 13:20', name='banana', qty=10, price=5.0)]
EDIT:
编辑：
New DataFramewas created with column names and old values was added. Last write to_csv:
DataFrame使用列名创建了新值，并添加了旧值。最后写to_csv：
df = df.set_index('datetime').astype(str)
df1 = pd.DataFrame(np.tile(np.array(df.columns), len(df.index)).reshape(len(df.index), -1),
index=df.index,
columns=df.columns) + '='
df1 = df1.add(df)
print (df1)
name qty price
datetime
2017-11-01 10:20 name=apple qty=5 price=1.0
2017-11-01 11:20 name=pear qty=2 price=1.5
2017-11-01 13:20 name=banana qty=10 price=5.0
df1.to_csv('filename.csv', header=None)
2017-11-01 10:20,name=apple,qty=5,price=1.0
2017-11-01 11:20,name=pear,qty=2,price=1.5
2017-11-01 13:20,name=banana,qty=10,price=5.0
回答by languitar
If you are happy with a dictionary as output, you can use
如果您对字典作为输出感到满意，则可以使用
df.to_dict('index')
On your example (with a slight parsing error for the dates by read_clipboard) this results in:
在您的示例中（日期 by 有轻微的解析错误read_clipboard）这导致：
In [17]: df = pd.read_clipboard().reset_index(drop=True)
In [18]: df.to_dict('index')
Out[18]:
{0: {'datetime': '10:20', 'name': 'apple', 'price': 1.0, 'qty': 5},
1: {'datetime': '11:20', 'name': 'pear', 'price': 1.5, 'qty': 2},
2: {'datetime': '13:20', 'name': 'banana', 'price': 5.0, 'qty': 10}}
回答by Charan P
If you exactly need the output you specified one possible solution is to iterate over the dataframe by rows and construct a string by zipping columns list with row values. A crude version is:
如果您确实需要您指定的输出，则一种可能的解决方案是按行迭代数据帧，并通过使用行值压缩列列表来构造一个字符串。一个粗略的版本是：
def create_key_vals(iterable):
str_list = []
for ite in iterable:
str_list.append(ite[0]+"="+ str(ite[1]))
return ','.join(str_list)
new_df = df.drop('datetime', axis=1)
col_list = new_df.columns.tolist()
zipped_col_vals = [zip(col_list, row) for row in new_df.itertuples(index=False, name=False)]
col_value_list = [create_key_vals(ite) for ite in zipped_col_vals]
In [116]: col_value_list
Out[116]:
['name=apple,price=1.0,qty=5',
'name=pear,price=1.5,qty=2',
'name=banana,price=5.0,qty=10']
You can convert this into a string by:
您可以通过以下方式将其转换为字符串：
In [117]: ('\n').join(col_value_list)
Out[117]: 'name=apple,price=1.0,qty=5\nname=pear,price=1.5,qty=2\nname=banana,price=5.0,qty=10
