Iterativelly Calculate in Separate Array in Loop

Question

I have to arrays, one for predictions and another for true values.

Predictions:

array([[ 0.01365575,  0.01523909],
   [-0.00044139,  0.00269908],
   [ 0.03240077,  0.02496629],
   [ 0.03238605,  0.03045709],
   [ 0.03226778,  0.02878774],
   [ 0.03238199,  0.03221421]])

Real values:

array([[0.01212121, 0.01529052],
   [0.        , 0.0030581 ],
   [0.01818182, 0.01559633],
   [0.00757576, 0.007263  ],
   [0.00757576, 0.00382263],
   [0.00757576, 0.01070336]])

I am trying to calculate the std of the mae and rmse with the formula:

std_error = (1/n * sum(error_i- mean_error)^2)^1/2

So far I am trying to create an array with the mae and rmse values incrementally but with no success. I am implementing this:

for x in range(len(preds)):
    mae_std = (preds[:,0] - trues[['t1']])/x

for x in range(len(preds)):
    rmse_std = (((trues[['t2']] - preds[:,1])**2)/x)**1/2

This way just takes forever and never really ends not sure why. I expected the results to be an array with the incremental values of the error and then I could try to use them on the std_error formula. What am I doing wrong? Is there a way to achieve what I describe faster?

you can index the arrays directly and compute the same? something like np.mean((preds[:,1] - trues[:,1])**2) (just an e.g) or you can use scikit learn metric submodule as well. scikit-learn.org/stable/modules/generated/… — Aditya
– Aditya, Commented May 27, 2021 at 9:33

score 1 · Accepted Answer · 2021-05-27 09:50:39Z

1

Assuming your 2 arrays are modelizing (x,y) coordinates, you can compute the mse / mae as follows:

# compute MSE
mse = np.square(preds - real)
mse = np.sum(mse, axis = 1) / mse.shape[1]

# compute MAE
mae = np.sum(np.abs(preds - real), axis = 1) / mse.shape[1]

You can adapt your specific formula to match these...

edited May 27, 2021 at 9:50

answered May 27, 2021 at 9:37

user14977424

Sign up to request clarification or add additional context in comments.

2 Comments

Peter Over a year ago

This return a number. Shouldn't it return an array of the error values for each sample? Because I then want to put them in error_i on the std_error formula.

user14977424 Over a year ago

if your initial pred & real arrays are of size (n, 2) then this returns an array of size (n, 1)

Collectives™ on Stack Overflow

Iterativelly Calculate in Separate Array in Loop

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related