How to read element by element an array inside an array in python

Question

I want to calculate the MSE of between two vectors, the original one and a prediction. X_train is a vector with all the original inputs and reconstructed32 are all the predictions, but I only want to calculate the MSE for the first element (X_train[0] and its prediction reconstructed32[0]), and I want do do it manually using some sort of function as the following:

mse=[]
for coef, coef32 in zip(X_train[0], reconstructed32[0]):
    mse.append((coef-coef32)**2)
print("MSE is", np.mean(mse))

But coef and coef32 do not access all the coefficients of the vector, instead they take the full vector and the for does only one iteration. This is how X_train[0] looks like:

array([[0.00000000e+00, 0.00000000e+00, 2.26577651e-02, 4.44901595e-03,
        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
        1.19896736e-02, 2.88007129e-03, 0.00000000e+00, 0.00000000e+00,
        1.89431056e-01, 0.00000000e+00, 3.64951454e-02, 0.00000000e+00,
        3.19024414e-01, 7.74396257e-03, 6.00360870e-01, 0.00000000e+00,
        6.25879224e-03, 6.58161461e-01, 8.55771124e-01, 0.00000000e+00,
        9.72116515e-02, 1.20666134e-03, 1.45429194e-01, 0.00000000e+00,
        8.25748563e-01, 6.86123502e-03, 5.37337780e-01, 0.00000000e+00,
        0.00000000e+00, 5.24137542e-03, 0.00000000e+00, 3.74730385e-04,
        0.00000000e+00, 1.18224040e-01, 8.56024176e-02, 0.00000000e+00,
        2.49160156e-02, 0.00000000e+00, 8.91805351e-01, 2.61418521e-03,
        0.00000000e+00, 1.73141190e-03, 0.00000000e+00, 4.95929737e-04,
        0.00000000e+00, 1.00000000e+00, 4.93207388e-02, 5.37037617e-03,
        3.81159922e-03, 2.94345170e-01, 4.77780886e-02, 0.00000000e+00,
        5.86739518e-02, 1.02838585e-02, 1.41941339e-01, 3.23100435e-03,
        0.00000000e+00, 1.34332031e-01, 2.35870923e-03, 0.00000000e+00,
        5.09367557e-03, 0.00000000e+00, 1.07753032e-03, 0.00000000e+00,
        0.00000000e+00, 0.00000000e+00, 3.47328931e-01, 0.00000000e+00,
        0.00000000e+00, 2.65773922e-01, 1.57823116e-02, 5.41949784e-03,
        7.56190493e-05, 1.40293926e-01, 5.02460590e-03, 6.51085284e-03,
        9.38402303e-03, 2.63461888e-01, 1.96186375e-04, 1.39610067e-01,
        0.00000000e+00, 2.39743218e-02, 3.54982950e-02, 6.54916763e-02,
        5.11557400e-01, 3.56561318e-03, 5.72423302e-02, 0.00000000e+00,
        1.21502653e-01, 2.71597654e-02, 2.47318600e-03, 6.85532205e-03,
        5.31266391e-01, 0.00000000e+00, 0.00000000e+00, 2.94529963e-02,
        0.00000000e+00, 3.75285745e-01, 3.31357837e-01, 1.18085509e-03,
        1.19613800e-02, 6.72158494e-05, 0.00000000e+00, 0.00000000e+00,
        0.00000000e+00, 7.86180317e-01, 0.00000000e+00, 0.00000000e+00,
        4.27392730e-03, 2.27557942e-02, 1.35929761e-02, 1.27607910e-02,
        0.00000000e+00, 2.63269083e-03, 1.48336608e-02, 0.00000000e+00,
        1.77292936e-02, 1.18409727e-04, 0.00000000e+00, 5.49420249e-03,
        0.00000000e+00, 1.70363311e-03, 1.75287843e-01, 4.07821295e-04]],
      dtype=float32)

As an example, if I try this simple code:

for coef in X_train[0]:
    print("coef is", coef)

I get this output:

coef is [0.00000000e+00 0.00000000e+00 2.26577651e-02 4.44901595e-03
 0.00000000e+00 0.00000000e+00 0.00000000e+00 0.00000000e+00
 1.19896736e-02 2.88007129e-03 0.00000000e+00 0.00000000e+00
 1.89431056e-01 0.00000000e+00 3.64951454e-02 0.00000000e+00
 3.19024414e-01 7.74396257e-03 6.00360870e-01 0.00000000e+00
 6.25879224e-03 6.58161461e-01 8.55771124e-01 0.00000000e+00
 9.72116515e-02 1.20666134e-03 1.45429194e-01 0.00000000e+00
 8.25748563e-01 6.86123502e-03 5.37337780e-01 0.00000000e+00
 0.00000000e+00 5.24137542e-03 0.00000000e+00 3.74730385e-04
 0.00000000e+00 1.18224040e-01 8.56024176e-02 0.00000000e+00
 2.49160156e-02 0.00000000e+00 8.91805351e-01 2.61418521e-03
 0.00000000e+00 1.73141190e-03 0.00000000e+00 4.95929737e-04
 0.00000000e+00 1.00000000e+00 4.93207388e-02 5.37037617e-03
 3.81159922e-03 2.94345170e-01 4.77780886e-02 0.00000000e+00
 5.86739518e-02 1.02838585e-02 1.41941339e-01 3.23100435e-03
 0.00000000e+00 1.34332031e-01 2.35870923e-03 0.00000000e+00
 5.09367557e-03 0.00000000e+00 1.07753032e-03 0.00000000e+00
 0.00000000e+00 0.00000000e+00 3.47328931e-01 0.00000000e+00
 0.00000000e+00 2.65773922e-01 1.57823116e-02 5.41949784e-03
 7.56190493e-05 1.40293926e-01 5.02460590e-03 6.51085284e-03
 9.38402303e-03 2.63461888e-01 1.96186375e-04 1.39610067e-01
 0.00000000e+00 2.39743218e-02 3.54982950e-02 6.54916763e-02
 5.11557400e-01 3.56561318e-03 5.72423302e-02 0.00000000e+00
 1.21502653e-01 2.71597654e-02 2.47318600e-03 6.85532205e-03
 5.31266391e-01 0.00000000e+00 0.00000000e+00 2.94529963e-02
 0.00000000e+00 3.75285745e-01 3.31357837e-01 1.18085509e-03
 1.19613800e-02 6.72158494e-05 0.00000000e+00 0.00000000e+00
 0.00000000e+00 7.86180317e-01 0.00000000e+00 0.00000000e+00
 4.27392730e-03 2.27557942e-02 1.35929761e-02 1.27607910e-02
 0.00000000e+00 2.63269083e-03 1.48336608e-02 0.00000000e+00
 1.77292936e-02 1.18409727e-04 0.00000000e+00 5.49420249e-03
 0.00000000e+00 1.70363311e-03 1.75287843e-01 4.07821295e-04]

And I would like to get "coef is" in front of every coefficient, not in front of the full vector. How can I solve this?

Thanks!

Answer 1

When you iterate through a NumPy array, you are iterating through the first index (i.e. the rows). Since you have a matrix of dimension (1, d), so iterating gives you the whole first row.

A few ways you can fix this, but the easiest way is just to do it with the row vector itself (of dimension (d,)).

# in your case, your_idx is 0
for coef, coef32 in zip(X_train[your_idx][0], reconstructed32[your_idx][0]):
   ... # do your thing

Answer 2

try this,

for coef in X_train[0][0]:
    print("coef is", coef)