从 Pandas 数据框列单独创建 numpy 数组

Question

I would like to create automatically in Python n of numpy arrays from my pandas dataframe columns. I can do this manually using for example:

numpy_array_1 = data_frame.column_1.values
numpy_array_2 = data_frame.column_2.values
...
numpy_array_n = data_frame.column_n.values

But I can not know how I should write code to create those arrays automatically.

Answer 1

You can simply use a for and loop through it. Remember that using (list(data_frame)) returns a list of the column names in the dataframe:

np_array = []
for i in list(data_frame):
    np_array.append(data_frame[i].values)

The expected output is a list that contains sublists of values. Where each sublist matches the position of the columns in the dataframe. Therefore you can either make a dictionary, or a tuple out of it. Dictionary example:

np_array_dict = {}
for i in list(data_frame):
    np_array_dict[i] = data_frame[i].values

Answer 2

Suppose we have a simple df:

df = pd.DataFrame({"0":[1,2], "1":[3,4]})
df 
   0  1
0  1  3
1  2  4

Then you can run:

for (key,value) in df.to_dict("list").items():
    exec("numpy_array_{} = np.array({})".format(key, value))

You'll get:

numpy_array_0
array([1, 2])

numpy_array_1
array([3, 4])

and so on.

Alternatively:

for col in list(df):
    exec("numpy_array_{} = df[str({})].values".format(col,col))

Answer 3

您可以获得所有数据帧行和列值的矩阵，就像 df.values 一样简单您真的需要每列不同的数组吗？

Answer 4

This can be done without using loops:

df = pd.DataFrame({"0":[1,2], "1":[3,4], "2":[5,6]})
print(df)

   0  1  2
0  1  3  5
1  2  4  6

and then:

[*np.transpose(df.values)]

results in:

[array([1, 2]), array([3, 4]), array([5, 6])]

and if a dictionary is desired one just needs to proceed as follows:

dict(zip(range(df.shape[1]), [*np.transpose(df.values)]))

which gives:

{0: array([1, 2]), 1: array([3, 4]), 2: array([5, 6])}