Merging straightening and remodeling instances of arrays in numpy and pandas

Merge

Merging two array in numpy

In numpy, concatenate can be used, and the parameter axis=0 indicates that two arrays are merged in the vertical direction, which is equivalent to np. vstack; The parameter axis=1 indicates that two arrays are merged horizontally, which is equivalent to np. hstack.

Vertical direction:

np.concatenate([arr1,arr2],axis=0)
np.vstack([arr1,arr2])

Horizontal direction:

np.concatenate([arr1,arr2],axis=1)
np.hstack([arr1,arr2])

import numpy as np
import pandas as pd

arr1=np.ones((3,5))
arr1
Out[5]: 
array([[ 1., 1., 1., 1., 1.],
    [ 1., 1., 1., 1., 1.],
    [ 1., 1., 1., 1., 1.]])
    
arr2=np.random.randn(15).reshape(arr1.shape)
arr2
Out[8]: 
array([[-0.09666833, 1.47064828, -1.94608976, 0.2651279 , -0.32894787],
    [ 1.01187699, 0.39171167, 1.49607091, 0.79216196, 0.33246644],
    [ 1.71266238, 0.86650837, 0.77830394, -0.90519422, 1.55410056]])
    
np.concatenate([arr1,arr2],axis=0) # Merge on the vertical axis 
Out[9]: 
array([[ 1.    , 1.    , 1.    , 1.    , 1.    ],
    [ 1.    , 1.    , 1.    , 1.    , 1.    ],
    [ 1.    , 1.    , 1.    , 1.    , 1.    ],
    [-0.09666833, 1.47064828, -1.94608976, 0.2651279 , -0.32894787],
    [ 1.01187699, 0.39171167, 1.49607091, 0.79216196, 0.33246644],
    [ 1.71266238, 0.86650837, 0.77830394, -0.90519422, 1.55410056]])

np.concatenate([arr1,arr2],axis=1) # Merge on the horizontal axis 
Out[10]: 
array([[ 1.    , 1.    , 1.    , ..., -1.94608976,
     0.2651279 , -0.32894787],
    [ 1.    , 1.    , 1.    , ..., 1.49607091,
     0.79216196, 0.33246644],
    [ 1.    , 1.    , 1.    , ..., 0.77830394,
    -0.90519422, 1.55410056]])
    
   
    

Merging two DataFrame in pandas

In pandas, the merge is implemented through the concat method, specifying the parameter axis=0 or axis=1, and merging two arrays on the vertical and horizontal axes. Unlike numpy, the two dataframe are placed in one list, namely [frame1, frame2]

from pandas import DataFrame

frame1=DataFrame([[1,2,3],[4,5,6]])
frame2=DataFrame([[7,8,9],[10,11,12]])

pd.concat([frame1,frame2], ignore_index=True) #  The merged array is 1 Iterable lists. 
Out[25]: 
  0  1  2
0  1  2  3
1  4  5  6
0  7  8  9
1 10 11 12


pd.concat([frame1,frame2], axis=1, ignore_index=True)
Out[27]: 
  0 1 2  3  4  5
0 1 2 3  7  8  9
1 4 5 6 10 11 12

Straightening and reshaping

Straightening means changing a 2-dimensional array into a 1-dimensional array. By default, Numpy arrays are created in row-first order. In terms of space, this means that for a 2-dimensional number, the data items in each row are stored in adjacent positions within. Another order is column first.

For historical reasons, row-first and column-first are also referred to as C and Fortran sequences, respectively. In Numpy, row and column precedence can be achieved by keyword parameters order= 'C' and order= 'F'.

Straighten:

arr=np.arange(15).reshape(3,-1)
arr
Out[29]: 
array([[ 0, 1, 2, 3, 4],
    [ 5, 6, 7, 8, 9],
    [10, 11, 12, 13, 14]])
    
arr.ravel('F') # Give priority to columns and flatten them. 
Out[30]: array([ 0, 5, 10, ..., 4, 9, 14])

arr.ravel('C') #  Default order. # According to the row priority, flat. 
Out[31]: array([ 0, 1, 2, ..., 12, 13, 14])

Remodeling:

After Fortran sequence remodeling, straighten by column and straighten by column the original data.

arr.reshape((5,3),order='F') 
Out[32]: 
array([[ 0, 11, 8],
    [ 5, 2, 13],
    [10, 7, 4],
    [ 1, 12, 9],
    [ 6, 3, 14]])

After C sequence remodeling, straighten by row and straighten by row the original data.

 arr.reshape((5,3),order='C')
 Out[33]: 
array([[ 0, 1, 2],
    [ 3, 4, 5],
    [ 6, 7, 8],
    [ 9, 10, 11],
    [12, 13, 14]])