Numpy array with position

Question

I have 14 array with a shape (25,43). To all the arrays I pass a mask

max=np.ma.masked_where(_mascara<0.5,max)
min=np.ma.masked_where(_mascara<0.5,min)

This mask create nan positions, that I will like discard.

I will like save in txt a join array with the position in x(0 to 25) and y(0 to 43) and the value (that is 14 values) but eliminate the mased positions to free size.

Example:

x=0,y=12,max,min...tilt

If you dont mind help me. Thank you and sorry my bad english.

CODE:

 m,n=np.mgrid[slice(0,25, 1),slice(0,43, 1)]


for x in variables:
    for y in dias:
        print y
        tmp=array(TODOS[y[0]:y[1],:,:,x])
        max=np.max(tmp,axis=0)
        max=np.ma.masked_where(_mascara<0.5,max)
        #max=np.ma.compressed(max)
        min=np.min(tmp,axis=0)
        min=np.ma.masked_where(_mascara<0.5,min)
        #min=np.ma.compressed(min)
        posicion=array(m)
        posicion=np.ma.masked_where(_mascara<0.5,m)
        #posicion=np.ma.compressed(posicion)

        print posicion.shape
        print max.shape
        print min.shape
        salida=array([m,max,min])
        np.savetxt('C:\\prueba_'+mapa[x]+'.csv',salida,delimiter=';')

I need (position,values) after this compresed and eliminate bad information

Have you considered something like ii = np.where(x>0.5) to obtain a list of indeces with the data you want and the generate a filtered array with x_new = x[ii]? — Dietrich
– Dietrich, Commented Feb 3, 2014 at 23:55

tom10 · Accepted Answer · 2014-02-04 19:21:55Z

1

In general you can use `MaskedArray.compressed, but for your specific requirements (to list the indices together with the data) you could do it all directly

>>> x = np.ma.array(100+np.arange(9), mask=[0,1,0,0,0,1,1,1,0]).reshape((3,3))

    masked_array(data =
    [[100 -- 102]
     [103 104 --]
     [-- -- 108]],
         mask =
    [[False  True False]
     [False False  True]
     [ True  True False]],
         fill_value = 999999)

>>> i = np.nonzero(~np.ma.getmask(x))   # get the indices of the unmasked items

    (array([0, 0, 1, 1, 2]), array([0, 2, 0, 1, 2]))

>>> np.vstack((i, x[i]))  # build an array with the indices and elements together

    [[  0   0   1   1   2]
     [  0   2   0   1   2]
     [100 102 103 104 108]]

or, if you just want the paired indices as tuples you could use:

>>> zip(*i)

    [(0, 0), (0, 2), (1, 0), (1, 1), (2, 2)]

I'll leave it at this, without further edits. With i and x[i] you can easily get anything you want (and if you want the masked items, just drop the ~ in the expression using nonzero).

edited Feb 4, 2014 at 19:21

answered Feb 3, 2014 at 23:56

tom10

69.5k11 gold badges133 silver badges143 bronze badges

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

8 Comments

virtualsets Over a year ago

Thank you for answer me.The problem is that I Lost the position x,y . I need have the postion to post the data after I need the position of the point (x,y) before compressed()

virtualsets Over a year ago

Thank you. The part of the mask I have clear the problem is save position before

tom10 Over a year ago

Sorry, but I don't understand how this isn't what you're asking for. Regardless, using the np.nonzero function you can get whatever indices you want: masked, unmasked, before, after, etc. I think you're asking to save the indices of the elements before they were compressed, and that's what my method above does.

virtualsets Over a year ago

Thank you. My English sucks. I Need before the value, its position in the array.example [[6,7],[8,9]] I need that save file 1 --> 0,0,6 file 2 -->0,1,7 file 3 -->1,0,8 file 4 --> 1,1,9

tom10 Over a year ago

OK, I've taken another shot at this. I hope this works for you.

|

Collectives™ on Stack Overflow

Numpy array with position

1 Answer 1

8 Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

8 Comments

Related