What does reshaping an array in Python with one dimension set to -1 do?

Aloha!!

According to numpy docs -

Once the shape dimension can be -1, the value is offered from the length of the array and remaining dimension.

Let's see how the value of -1 is calculated.

Eg -

1. import numpy as np 
2. z = np.array([ 
3. [1,2,3,4], 
4. [5,6,7,8], 
5. [9,10,11,12]]) 
6.  
7. z.shape=>(3,4) 
8. z.size => 12 =>3*4 
9. z.ndin=>2, z is 2d 

Now if .

1. z.reshape(-1)  
2. [ 1 2 3 4 5 6 7 8 9 10 11 12]  
3. z.reshape(-1).ndim => 1 

Comments - z has been reduced to 1d array which equals to size of original array z. So -1 == 12(size of original z).

Now if.

1. z.reshape((-1,1)) 
2. [[ 1] 
3. [ 2]  
4. [ 3]  
5. [ 4]  
6. [ 5]  
7. [ 6]  
8. [ 7]  
9. [ 8]  
10. [ 9]  
11. [10]  
12. [11]  
13. [12]]  

Comments -

1. Now shape of z is (12,1).
2. Passed value (-1,1)
3. Calculation of -1 = size of z /column = 12/1 = 12
4. So shape is (12,1).

Now if.

1. z.reshape((-1,2)) 
2. [[ 1 2] 
3. [ 3 4]  
4. [ 5 6]  
5. [ 7 8]  
6. [ 9 10]  
7. [11 12]]  

Comments -

1. Now shape of z is (6,2).
2. Passed value(-1,2).
3. Calculation of -1 = size of z /column = 12/2 = 6.
4. So the shape is (6,2).

Now if.

1. z.reshape((3,-1)) 
2. [[ 1 2 3 4] 
3. [ 5 6 7 8]  
4. [ 9 10 11 12]]  

Comments -

1. Now shape of z is (3 ,4).
2. Passed size (3,-1)
3. Calculation of -1 = size of z /row= 12/3 =4.
4. So the shape is (3,4).

Now if.

1. z.reshape((-1,-1))  
2. ValueError: can only specify one unknown dimension  

Comments -

1. Value error.
2. Passed size (-1,-1).
3. Calculation of -1 = 12/12 = 1
4. Shape(1,1 ). A size 12 array cannot be reduced to shape (1,1). So we get the Error.

JD.