在Python中将3d数组关联为n乘2d行

Question

Was just wondering if there was a simple way to flatten a 2d array into correlation rows. Ok so here's an example: I have n numbers of a 2d array as such:

So say example here a 3d (6x10x n) How can i correlate a1:a10, b1:b10 till f1:f10 so call them feature a1,a2,a3....f8,f9,f10 and do this for n rows. N rows of 60 features (a1,a2,a3...).

Answer for n rows is:

import numpy as np 
from numpy import array
from numpy import vstack
from numpy import hstack
import scipy.io
mat = scipy.io.loadmat('x.mat')

I've only been able to load the .mat file to python. Are there any functions i can use in Python? Thank you for any help.

Kevin

Answer 1

You can do something like this :

import numpy as np
import pandas as pd

Array = np.array(2d_array)
columns = Array[0, 1:]
rows = Array[1:, 0:1]
values = Array[1:,1:]

results = dict()
for row, column, value in np.nditer((rows, columns, values)):
    feature = "".join(map(str, [row, column]))
    if feature not in results:
       results[feature] = []
    results[feature].append(value)
df = pd.DataFrame.from_dict(results)