Given a list of df,the objective is to reshape the df_convert_array into 2D array.

The below snippet does fulfill the objective

alist_df=[pd.DataFrame(np.random.randint(100,size=(4,2))) for _ in range (3)]
arr=np.array(alist_df)
arr_opt=np.array([narr.T.reshape(-1) for narr in arr])

But, the proposed solution require to first transpose and reshape each array in the list.

I thought, the process can be simplified simply as below

arr_opt=arr.T.reshape(3,-1)

However, the output is not as I expect.

May I know how to avoid the use of list-comprehension as describe above?

Remark: Output should be in the form of 2D numpy array

CodePudding user response：

You can also transpose the array, reshape, and then transpose back:

arr_opt = np.reshape(arr.T, (arr.shape[2]*arr.shape[1], -1)).T

(the references to arr.shape are in case you don't know ahead of time what the dimensions of your dataframes are)

CodePudding user response：

In your second example, the transpose operation will permute all three axes before the final reshape, whereas in your first example, only the second two axes are effectively permuted.

Try arr_opt = arr.transpose((1,2)).reshape((3,-1))