CODEWITHSUNDEEP
pythonarrayspandasnumpy
humblefool_9
humblefool_9

Reputation: 89

Reshape a 3D array to a 2D array in Python

Suppose I have a data frame of length 5800000 which is a concatenation of 100 files where each file has 58000 rows. I have an array fv of shape (100, 10, 58000) which I want to add to the data frame by adding 10 columns. df has a length of 5800000 with two columns but only focuses on the first column index, i.e df.shape[0]

list_ = ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j']  
fv = np.zeros(len(data), len(list_), int(df.shape[0]/len(files))

def add_fv_to_dataframe(_data, list_):
    for index in range(len(_data):
        for name_index, name_in_list in enumerate(list_):
            calculate something
            calcs_ = _data[index]
            fv[index, name_index, :] = calcs_
            # add the calculated values to the dataframe

            df['fv_{}'.format(name_in_list)] = pd.Series(fv.reshape(-1, (10,1)), index=df.index)

I would like to have my final data frame in the form;

df[0] df[1] fv_a fv_b fv_c fv_d fv_e fv_f fv_g fv_h fv_i fv_j
1 1 1 1 1 1 1 1 1 1 1 1
: : : : : : : : : : : :
5800000 5800000 5800000 5800000 5800000 5800000 5800000 5800000 5800000 5800000 5800000 5800000

Upvotes: 1

Views: 71

Answers (1)

Code Different
Code Different

Reputation: 93191

Use np.swapaxes:

for i, data in enumerate(np.swapaxes(fv, 0, 1)):
    df[f"fv_{i}"] = np.ravel(data)

Upvotes: 1

Related Questions