ArcAngel - 8 months ago 32

Python Question

I’m currently writing something that involves a lot of noise I’m attempting to remove, but in order do this I initially used masks, but the way in which I’m analysing the data breaks using a mask.

The masking is done, I’m looking to extract the data that is not masked, run analysis on this, then rebuild the array with the original order.

`array([[3, 0, 3],`

[6, 7, 2],

[2, 5, 0],

[2, 1, 4]])

Make Mask

`array([[-, -, -],`

[6, 7, 2],

[-, -, -],

[2, 1, 4]])

Extract Values

`array([[6, 7, 2],`

[2, 1, 4]])

Do analysis

Rebuild Array

`array([[-, -, -],`

[6, 7, 2],

[-, -, -],

[2, 1, 4]])

I’m hoping for an efficient way of doing this as I’m dealing with 100 million data points. Any suggestions are appreciated.

Answer

You could use

```
masked[~masked.mask] = analyzed.ravel()
```

to reassign the analyzed values to the masked array.

```
import numpy as np
arr = np.array([[3, 0, 3],
[6, 7, 2],
[2, 5, 0],
[2, 1, 4]])
masked = np.ma.masked_array(arr, mask=False)
masked.mask[::2, None] = True
extracted = np.ma.compress_rows(masked)
analyzed = extracted*10
masked[~masked.mask] = analyzed.ravel()
print(masked)
```

yields

```
[[-- -- --]
[60 70 20]
[-- -- --]
[20 10 40]]
```