How to create a pandas DataFrame with several numpy 1d arrays?

Question

I've created some np.arrays to do some calculation with them. (All have the same size [100,1]) Now I want to create a pandas Dataframe and each array shoud be one column of that DF. The Names of the arrays should be the header of the DataFrame.

In Matlab I would easily do it like that:

Table = table(array1, array2, array3, ... );

How can I do this in Python?

thanks in advance!

Ah sorry I misread your question. You would need to reshape the arrays to use the syntax I provided. — Andrew L
– Andrew L, Commented Jul 30, 2017 at 13:12

score 9 · Accepted Answer · 2017-07-30 13:13:02Z

9

Let's say these are your arrays:

arr1, arr2, arr3 = np.zeros((3, 100, 1))

arr1.shape
Out: (100, 1)

You can use hstack to stack them and pass the resulting 2D array to the DataFrame constructor:

df = pd.DataFrame(np.hstack((arr1, arr2, arr3)))

df.head()
Out: 
     0    1    2
0  0.0  0.0  0.0
1  0.0  0.0  0.0
2  0.0  0.0  0.0
3  0.0  0.0  0.0
4  0.0  0.0  0.0

Or name the columns as arr1, arr2, ...:

df = pd.DataFrame(np.hstack((arr1, arr2, arr3)), 
                  columns=['arr{}'.format(i+1) for i in range(3)])

which gives

df.head()
Out: 
   arr1  arr2  arr3
0   0.0   0.0   0.0
1   0.0   0.0   0.0
2   0.0   0.0   0.0
3   0.0   0.0   0.0
4   0.0   0.0   0.0

edited Jul 30, 2017 at 13:13

answered Jul 30, 2017 at 12:49

user2285236

Sign up to request clarification or add additional context in comments.

3 Comments

laurenz Over a year ago

Thank you, that worked well! Is it possible that the header gets the arrays names instead of 0,1,2 ?

user2285236 Over a year ago

In Python, names of the objects are just references. An object may have multiple names. Because of that there is no easy way of getting the name of a variable, say arr1. You can iterate over the names in globals() but it is not recommended. Do your variable names have a structure? If so, it is easy to build the names instead of extracting them. (Like arr1, arr2, arr3...)

laurenz Over a year ago

Thank you. No they are real names. So i think i really need to rename them in the Dataframe then.

jezrael · Accepted Answer · 2017-07-30 13:09:16Z

2

Solution with numpy.concatenate for 2d array and DataFrame constructor:

df = pd.DataFrame(np.concatenate([arr1, arr2, arr3], axis=1), columns= ['a','b','c'])

edited Jul 30, 2017 at 13:09

answered Jul 30, 2017 at 12:51

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

Collectives™ on Stack Overflow

How to create a pandas DataFrame with several numpy 1d arrays?

2 Answers 2

3 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related