DataArray

class Stoner.core.array.DataArray(input_array, *args, **kargs)[source]

Bases: numpy.ma.MaskedArray

A sub class of numpy.ma.MaskedArray with a copy of the setas attribute to allow indexing by name.

column_headers

of strings of the column names of the data.

Type

list

i

When read, returns the row umbers of the data. When written to, sets the base row index. The base row index is preserved when a DataArray is indexed.

Type

array of integers

x,y,z

When a column is declared to contain x, y, or z data, then these attributes access the corresponding columns. When written to, the attributes overwrite the existing column’s data.

Type

1D DataArray

d,e,f

Where a column is identified as containing uncertainities for x, y or z data, then these attributes provide a quick access to them. When written to, the attributes overwrite the existing column’s data.

Type

1D DataArray

u,v,w

Columns may be identieid as containing vectgor field information. These attributes provide quick access to them, assuming that they are defined as cartesian co-ordinates. When written to, the attributes overwrite the existing column’s data.

Type

1D DataArray

p,q,r

These attributes access calculated columns that convert \((x,y,z)\) data or \((u,v,w)\) into \((\phi,\theta,r)\) polar co-ordinates. If on x and y columns are defined, then 2D polar co-ordinates are returned for q and r.

Type

1D DataArray

setas

Actually a proxy to a magic class that handles the assignment of columns to different axes and also tracks the names of columns (so that columns may be accessed as named items).

Type

list or string

This array type is used to represent numeric data in the Stoner Package - primarily as a 2D matrix in Stoner.Core.DataFile but also when a 1D row is required. In con trast to the parent class, DataArray understands that it came from a DataFile which has a setas attribute and column assignments. This allows the row to be indexed by column name, and also for quick attribute access to work. This makes writing functions to work with a single row of data more attractive.

Attributes Summary

T

The transposed array.

base

Base object if memory is from some other object.

baseclass

Class of the underlying data (read-only).

column_headers

Pass through to the setas attribute.

ctypes

An object to simplify the interaction of the array with the ctypes module.

data

Returns the underlying data, as a view of the masked array.

dtype

Data-type of the array's elements.

fill_value

The filling value of the masked array is a scalar.

flags

Information about the memory layout of the array.

flat

Return a flat iterator, or set a flattened version of self to value.

hardmask

Hardness of the mask

i

Return the row indices of the DataArray or sets the base index - the row number of the first row.

imag

The imaginary part of the masked array.

isrow

Define whether this is a single row or a column if 1D.

itemsize

Length of one array element in bytes.

mask

Current mask.

nbytes

Total bytes consumed by the elements of the array.

ndim

Number of array dimensions.

p

Calculate the inclination \(\phi\) co-ordinate for spherical co-ordinate systems.

q

Calculate the azimuthal \(\theta\) co-ordinate if using spherical or polar co-ordinates.

r

Calculate the radius \(\rho\) co-ordinate if using spherical or polar co-ordinate systems.

real

The real part of the masked array.

recordmask

Get or set the mask of the array if it has no named fields.

setas

Return an object for setting column assignments.

shape

Tuple of array dimensions.

sharedmask

Share status of the mask (read-only).

size

Number of elements in the array.

strides

Tuple of bytes to step in each dimension when traversing an array.

Methods Summary

all([axis, out, keepdims])

Returns True if all elements evaluate to True.

anom([axis, dtype])

Compute the anomalies (deviations from the arithmetic mean) along the given axis.

any([axis, out, keepdims])

Returns True if any of the elements of a evaluate to True.

argmax([axis, fill_value, out])

Returns array of indices of the maximum values along the given axis.

argmin([axis, fill_value, out])

Return array of indices to the minimum values along the given axis.

argpartition(kth[, axis, kind, order])

Returns the indices that would partition this array.

argsort([axis, kind, order, endwith, fill_value])

Return an ndarray of indices that sort the array along the specified axis.

astype(dtype[, order, casting, subok, copy])

Copy of the array, cast to a specified type.

byteswap([inplace])

Swap the bytes of the array elements

choose(choices[, out, mode])

Use an index array to construct a new array from a set of choices.

clip([min, max, out])

Return an array whose values are limited to [min, max].

compress(condition[, axis, out])

Return a where condition is True.

compressed()

Return all the non-masked data as a 1-D array.

conj()

Complex-conjugate all elements.

conjugate()

Return the complex conjugate, element-wise.

copy([order])

Return a copy of the array.

count([axis, keepdims])

Count the non-masked elements of the array along the given axis.

cumprod([axis, dtype, out])

Return the cumulative product of the array elements over the given axis.

cumsum([axis, dtype, out])

Return the cumulative sum of the array elements over the given axis.

diagonal([offset, axis1, axis2])

Return specified diagonals.

dot(b[, out])

Masked dot product of two arrays.

dump(file)

Dump a pickle of the array to the specified file.

dumps()

Returns the pickle of the array as a string.

fill(value)

Fill the array with a scalar value.

filled([fill_value])

Return a copy of self, with masked values filled with a given value.

flatten([order])

Return a copy of the array collapsed into one dimension.

get_fill_value()

The filling value of the masked array is a scalar.

get_imag()

The imaginary part of the masked array.

get_real()

The real part of the masked array.

getfield(dtype[, offset])

Returns a field of the given array as a certain type.

harden_mask()

Force the mask to hard.

ids()

Return the addresses of the data and mask areas.

iscontiguous()

Return a boolean indicating whether the data is contiguous.

item(*args)

Copy an element of an array to a standard Python scalar and return it.

itemset(*args)

Insert scalar into an array (scalar is cast to array's dtype, if possible)

keys()

Return a list of column headers.

max([axis, out, fill_value, keepdims])

Return the maximum along a given axis.

mean([axis, dtype, out, keepdims])

Returns the average of the array elements along given axis.

min([axis, out, fill_value, keepdims])

Return the minimum along a given axis.

mini([axis])

Return the array minimum along the specified axis.

newbyteorder([new_order])

Return the array with the same data viewed with a different byte order.

nonzero()

Return the indices of unmasked elements that are not zero.

partition(kth[, axis, kind, order])

Rearranges the elements in the array in such a way that the value of the element in kth position is in the position it would be in a sorted array.

prod([axis, dtype, out, keepdims])

Return the product of the array elements over the given axis.

product([axis, dtype, out, keepdims])

Return the product of the array elements over the given axis.

ptp([axis, out, fill_value, keepdims])

Return (maximum - minimum) along the given dimension (i.e.

put(indices, values[, mode])

Set storage-indexed locations to corresponding values.

ravel([order])

Returns a 1D version of self, as a view.

repeat(repeats[, axis])

Repeat elements of an array.

reshape(*s, **kwargs)

Give a new shape to the array without changing its data.

resize(newshape[, refcheck, order])

round([decimals, out])

Return each element rounded to the given number of decimals.

searchsorted(v[, side, sorter])

Find indices where elements of v should be inserted in a to maintain order.

set_fill_value([value])

setfield(val, dtype[, offset])

Put a value into a specified place in a field defined by a data-type.

setflags([write, align, uic])

Set array flags WRITEABLE, ALIGNED, (WRITEBACKIFCOPY and UPDATEIFCOPY), respectively.

shrink_mask()

Reduce a mask to nomask when possible.

soften_mask()

Force the mask to soft.

sort([axis, kind, order, endwith, fill_value])

Sort the array, in-place

squeeze([axis])

Remove axes of length one from a.

std([axis, dtype, out, ddof, keepdims])

Returns the standard deviation of the array elements along given axis.

sum([axis, dtype, out, keepdims])

Return the sum of the array elements over the given axis.

swap_column(*swp, **kargs)

Swap pairs of columns in the data.

swapaxes(axis1, axis2)

Return a view of the array with axis1 and axis2 interchanged.

take(indices[, axis, out, mode])

tobytes([fill_value, order])

Return the array data as a string containing the raw bytes in the array.

tofile(fid[, sep, format])

Save a masked array to a file in binary format.

toflex()

Transforms a masked array into a flexible-type array.

tolist([fill_value])

Return the data portion of the masked array as a hierarchical Python list.

torecords()

Transforms a masked array into a flexible-type array.

tostring([fill_value, order])

A compatibility alias for tobytes, with exactly the same behavior.

trace([offset, axis1, axis2, dtype, out])

Return the sum along diagonals of the array.

transpose(*axes)

Returns a view of the array with axes transposed.

unshare_mask()

Copy the mask and set the sharedmask flag to False.

var([axis, dtype, out, ddof, keepdims])

Compute the variance along the specified axis.

view([dtype, type, fill_value])

Return a view of the MaskedArray data.

Attributes Documentation

T
base

Base object if memory is from some other object.

The base of an array that owns its memory is None:

>>> x = np.array([1,2,3,4])
>>> x.base is None
True

Slicing creates a view, whose memory is shared with x:

>>> y = x[2:]
>>> y.base is x
True
baseclass

Class of the underlying data (read-only).

column_headers

Pass through to the setas attribute.

ctypes

An object to simplify the interaction of the array with the ctypes module.

This attribute creates an object that makes it easier to use arrays when calling shared libraries with the ctypes module. The returned object has, among others, data, shape, and strides attributes (see Notes below) which themselves return ctypes objects that can be used as arguments to a shared library.

None

cPython object

Possessing attributes data, shape, strides, etc.

numpy.ctypeslib

Below are the public attributes of this object which were documented in “Guide to NumPy” (we have omitted undocumented public attributes, as well as documented private attributes):

_ctypes.data

A pointer to the memory area of the array as a Python integer. This memory area may contain data that is not aligned, or not in correct byte-order. The memory area may not even be writeable. The array flags and data-type of this array should be respected when passing this attribute to arbitrary C-code to avoid trouble that can include Python crashing. User Beware! The value of this attribute is exactly the same as self._array_interface_['data'][0].

Note that unlike data_as, a reference will not be kept to the array: code like ctypes.c_void_p((a + b).ctypes.data) will result in a pointer to a deallocated array, and should be spelt (a + b).ctypes.data_as(ctypes.c_void_p)

_ctypes.shape

A ctypes array of length self.ndim where the basetype is the C-integer corresponding to dtype('p') on this platform. This base-type could be ctypes.c_int, ctypes.c_long, or ctypes.c_longlong depending on the platform. The c_intp type is defined accordingly in numpy.ctypeslib. The ctypes array contains the shape of the underlying array.

Type

(c_intp*self.ndim)

_ctypes.strides

A ctypes array of length self.ndim where the basetype is the same as for the shape attribute. This ctypes array contains the strides information from the underlying array. This strides information is important for showing how many bytes must be jumped to get to the next element in the array.

Type

(c_intp*self.ndim)

_ctypes.data_as(obj)

Return the data pointer cast to a particular c-types object. For example, calling self._as_parameter_ is equivalent to self.data_as(ctypes.c_void_p). Perhaps you want to use the data as a pointer to a ctypes array of floating-point data: self.data_as(ctypes.POINTER(ctypes.c_double)).

The returned pointer will keep a reference to the array.

_ctypes.shape_as(obj)

Return the shape tuple as an array of some other c-types type. For example: self.shape_as(ctypes.c_short).

_ctypes.strides_as(obj)

Return the strides tuple as an array of some other c-types type. For example: self.strides_as(ctypes.c_longlong).

If the ctypes module is not available, then the ctypes attribute of array objects still returns something useful, but ctypes objects are not returned and errors may be raised instead. In particular, the object will still have the as_parameter attribute which will return an integer equal to the data attribute.

>>> import ctypes
>>> x = np.array([[0, 1], [2, 3]], dtype=np.int32)
>>> x
array([[0, 1],
       [2, 3]], dtype=int32)
>>> x.ctypes.data
31962608 # may vary
>>> x.ctypes.data_as(ctypes.POINTER(ctypes.c_uint32))
<__main__.LP_c_uint object at 0x7ff2fc1fc200> # may vary
>>> x.ctypes.data_as(ctypes.POINTER(ctypes.c_uint32)).contents
c_uint(0)
>>> x.ctypes.data_as(ctypes.POINTER(ctypes.c_uint64)).contents
c_ulong(4294967296)
>>> x.ctypes.shape
<numpy.core._internal.c_long_Array_2 object at 0x7ff2fc1fce60> # may vary
>>> x.ctypes.strides
<numpy.core._internal.c_long_Array_2 object at 0x7ff2fc1ff320> # may vary
data

Returns the underlying data, as a view of the masked array.

If the underlying data is a subclass of numpy.ndarray, it is returned as such.

>>> x = np.ma.array(np.matrix([[1, 2], [3, 4]]), mask=[[0, 1], [1, 0]])
>>> x.data
matrix([[1, 2],
        [3, 4]])

The type of the data can be accessed through the baseclass attribute.

dtype
fill_value

The filling value of the masked array is a scalar. When setting, None will set to a default based on the data type.

>>> for dt in [np.int32, np.int64, np.float64, np.complex128]:
...     np.ma.array([0, 1], dtype=dt).get_fill_value()
...
999999
999999
1e+20
(1e+20+0j)
>>> x = np.ma.array([0, 1.], fill_value=-np.inf)
>>> x.fill_value
-inf
>>> x.fill_value = np.pi
>>> x.fill_value
3.1415926535897931 # may vary

Reset to default:

>>> x.fill_value = None
>>> x.fill_value
1e+20
flags

Information about the memory layout of the array.

C_CONTIGUOUS (C)

The data is in a single, C-style contiguous segment.

F_CONTIGUOUS (F)

The data is in a single, Fortran-style contiguous segment.

OWNDATA (O)

The array owns the memory it uses or borrows it from another object.

WRITEABLE (W)

The data area can be written to. Setting this to False locks the data, making it read-only. A view (slice, etc.) inherits WRITEABLE from its base array at creation time, but a view of a writeable array may be subsequently locked while the base array remains writeable. (The opposite is not true, in that a view of a locked array may not be made writeable. However, currently, locking a base object does not lock any views that already reference it, so under that circumstance it is possible to alter the contents of a locked array via a previously created writeable view onto it.) Attempting to change a non-writeable array raises a RuntimeError exception.

ALIGNED (A)

The data and all elements are aligned appropriately for the hardware.

WRITEBACKIFCOPY (X)

This array is a copy of some other array. The C-API function PyArray_ResolveWritebackIfCopy must be called before deallocating to the base array will be updated with the contents of this array.

UPDATEIFCOPY (U)

(Deprecated, use WRITEBACKIFCOPY) This array is a copy of some other array. When this array is deallocated, the base array will be updated with the contents of this array.

FNC

F_CONTIGUOUS and not C_CONTIGUOUS.

FORC

F_CONTIGUOUS or C_CONTIGUOUS (one-segment test).

BEHAVED (B)

ALIGNED and WRITEABLE.

CARRAY (CA)

BEHAVED and C_CONTIGUOUS.

FARRAY (FA)

BEHAVED and F_CONTIGUOUS and not C_CONTIGUOUS.

The flags object can be accessed dictionary-like (as in a.flags['WRITEABLE']), or by using lowercased attribute names (as in a.flags.writeable). Short flag names are only supported in dictionary access.

Only the WRITEBACKIFCOPY, UPDATEIFCOPY, WRITEABLE, and ALIGNED flags can be changed by the user, via direct assignment to the attribute or dictionary entry, or by calling ndarray.setflags.

The array flags cannot be set arbitrarily:

  • UPDATEIFCOPY can only be set False.

  • WRITEBACKIFCOPY can only be set False.

  • ALIGNED can only be set True if the data is truly aligned.

  • WRITEABLE can only be set True if the array owns its own memory or the ultimate owner of the memory exposes a writeable buffer interface or is a string.

Arrays can be both C-style and Fortran-style contiguous simultaneously. This is clear for 1-dimensional arrays, but can also be true for higher dimensional arrays.

Even for contiguous arrays a stride for a given dimension arr.strides[dim] may be arbitrary if arr.shape[dim] == 1 or the array has no elements. It does not generally hold that self.strides[-1] == self.itemsize for C-style contiguous arrays or self.strides[0] == self.itemsize for Fortran-style contiguous arrays is true.

flat

Return a flat iterator, or set a flattened version of self to value.

hardmask

Hardness of the mask

i

Return the row indices of the DataArray or sets the base index - the row number of the first row.

imag

The imaginary part of the masked array.

This property is a view on the imaginary part of this MaskedArray.

real

>>> x = np.ma.array([1+1.j, -2j, 3.45+1.6j], mask=[False, True, False])
>>> x.imag
masked_array(data=[1.0, --, 1.6],
             mask=[False,  True, False],
       fill_value=1e+20)
isrow

Define whether this is a single row or a column if 1D.

itemsize

Length of one array element in bytes.

>>> x = np.array([1,2,3], dtype=np.float64)
>>> x.itemsize
8
>>> x = np.array([1,2,3], dtype=np.complex128)
>>> x.itemsize
16
mask

Current mask.

nbytes

Total bytes consumed by the elements of the array.

Does not include memory consumed by non-element attributes of the array object.

>>> x = np.zeros((3,5,2), dtype=np.complex128)
>>> x.nbytes
480
>>> np.prod(x.shape) * x.itemsize
480
ndim

Number of array dimensions.

>>> x = np.array([1, 2, 3])
>>> x.ndim
1
>>> y = np.zeros((2, 3, 4))
>>> y.ndim
3
p

Calculate the inclination \(\phi\) co-ordinate for spherical co-ordinate systems.

q

Calculate the azimuthal \(\theta\) co-ordinate if using spherical or polar co-ordinates.

r

Calculate the radius \(\rho\) co-ordinate if using spherical or polar co-ordinate systems.

real

The real part of the masked array.

This property is a view on the real part of this MaskedArray.

imag

>>> x = np.ma.array([1+1.j, -2j, 3.45+1.6j], mask=[False, True, False])
>>> x.real
masked_array(data=[1.0, --, 3.45],
             mask=[False,  True, False],
       fill_value=1e+20)
recordmask

Get or set the mask of the array if it has no named fields. For structured arrays, returns a ndarray of booleans where entries are True if all the fields are masked, False otherwise:

>>> x = np.ma.array([(1, 1), (2, 2), (3, 3), (4, 4), (5, 5)],
...         mask=[(0, 0), (1, 0), (1, 1), (0, 1), (0, 0)],
...        dtype=[('a', int), ('b', int)])
>>> x.recordmask
array([False, False,  True, False, False])
setas

Return an object for setting column assignments.

shape
sharedmask

Share status of the mask (read-only).

size

Number of elements in the array.

Equal to np.prod(a.shape), i.e., the product of the array’s dimensions.

a.size returns a standard arbitrary precision Python integer. This may not be the case with other methods of obtaining the same value (like the suggested np.prod(a.shape), which returns an instance of np.int_), and may be relevant if the value is used further in calculations that may overflow a fixed size integer type.

>>> x = np.zeros((3, 5, 2), dtype=np.complex128)
>>> x.size
30
>>> np.prod(x.shape)
30
strides

Tuple of bytes to step in each dimension when traversing an array.

The byte offset of element (i[0], i[1], ..., i[n]) in an array a is:

offset = sum(np.array(i) * a.strides)

A more detailed explanation of strides can be found in the “ndarray.rst” file in the NumPy reference guide.

Imagine an array of 32-bit integers (each 4 bytes):

x = np.array([[0, 1, 2, 3, 4],
              [5, 6, 7, 8, 9]], dtype=np.int32)

This array is stored in memory as 40 bytes, one after the other (known as a contiguous block of memory). The strides of an array tell us how many bytes we have to skip in memory to move to the next position along a certain axis. For example, we have to skip 4 bytes (1 value) to move to the next column, but 20 bytes (5 values) to get to the same position in the next row. As such, the strides for the array x will be (20, 4).

numpy.lib.stride_tricks.as_strided

>>> y = np.reshape(np.arange(2*3*4), (2,3,4))
>>> y
array([[[ 0,  1,  2,  3],
        [ 4,  5,  6,  7],
        [ 8,  9, 10, 11]],
       [[12, 13, 14, 15],
        [16, 17, 18, 19],
        [20, 21, 22, 23]]])
>>> y.strides
(48, 16, 4)
>>> y[1,1,1]
17
>>> offset=sum(y.strides * np.array((1,1,1)))
>>> offset/y.itemsize
17
>>> x = np.reshape(np.arange(5*6*7*8), (5,6,7,8)).transpose(2,3,1,0)
>>> x.strides
(32, 4, 224, 1344)
>>> i = np.array([3,5,2,2])
>>> offset = sum(i * x.strides)
>>> x[3,5,2,2]
813
>>> offset / x.itemsize
813

Methods Documentation

all(axis=None, out=None, keepdims=<no value>)

Returns True if all elements evaluate to True.

The output array is masked where all the values along the given axis are masked: if the output would have been a scalar and that all the values are masked, then the output is masked.

Refer to numpy.all for full documentation.

numpy.ndarray.all : corresponding function for ndarrays numpy.all : equivalent function

>>> np.ma.array([1,2,3]).all()
True
>>> a = np.ma.array([1,2,3], mask=True)
>>> (a.all() is np.ma.masked)
True
anom(axis=None, dtype=None)

Compute the anomalies (deviations from the arithmetic mean) along the given axis.

Returns an array of anomalies, with the same shape as the input and where the arithmetic mean is computed along the given axis.

axisint, optional

Axis over which the anomalies are taken. The default is to use the mean of the flattened array as reference.

dtypedtype, optional
Type to use in computing the variance. For arrays of integer type

the default is float32; for arrays of float types it is the same as the array type.

mean : Compute the mean of the array.

>>> a = np.ma.array([1,2,3])
>>> a.anom()
masked_array(data=[-1.,  0.,  1.],
             mask=False,
       fill_value=1e+20)
any(axis=None, out=None, keepdims=<no value>)

Returns True if any of the elements of a evaluate to True.

Masked values are considered as False during computation.

Refer to numpy.any for full documentation.

numpy.ndarray.any : corresponding function for ndarrays numpy.any : equivalent function

argmax(axis=None, fill_value=None, out=None)

Returns array of indices of the maximum values along the given axis. Masked values are treated as if they had the value fill_value.

axis{None, integer}

If None, the index is into the flattened array, otherwise along the specified axis

fill_valuescalar or None, optional

Value used to fill in the masked values. If None, the output of maximum_fill_value(self._data) is used instead.

out{None, array}, optional

Array into which the result can be placed. Its type is preserved and it must be of the right shape to hold the output.

index_array : {integer_array}

>>> a = np.arange(6).reshape(2,3)
>>> a.argmax()
5
>>> a.argmax(0)
array([1, 1, 1])
>>> a.argmax(1)
array([2, 2])
argmin(axis=None, fill_value=None, out=None)

Return array of indices to the minimum values along the given axis.

axis{None, integer}

If None, the index is into the flattened array, otherwise along the specified axis

fill_valuescalar or None, optional

Value used to fill in the masked values. If None, the output of minimum_fill_value(self._data) is used instead.

out{None, array}, optional

Array into which the result can be placed. Its type is preserved and it must be of the right shape to hold the output.

ndarray or scalar

If multi-dimension input, returns a new ndarray of indices to the minimum values along the given axis. Otherwise, returns a scalar of index to the minimum values along the given axis.

>>> x = np.ma.array(np.arange(4), mask=[1,1,0,0])
>>> x.shape = (2,2)
>>> x
masked_array(
  data=[[--, --],
        [2, 3]],
  mask=[[ True,  True],
        [False, False]],
  fill_value=999999)
>>> x.argmin(axis=0, fill_value=-1)
array([0, 0])
>>> x.argmin(axis=0, fill_value=9)
array([1, 1])
argpartition(kth, axis=- 1, kind='introselect', order=None)

Returns the indices that would partition this array.

Refer to numpy.argpartition for full documentation.

New in version 1.8.0.

numpy.argpartition : equivalent function

argsort(axis=<no value>, kind=None, order=None, endwith=True, fill_value=None)

Return an ndarray of indices that sort the array along the specified axis. Masked values are filled beforehand to fill_value.

axisint, optional

Axis along which to sort. If None, the default, the flattened array is used.

Changed in version 1.13.0: Previously, the default was documented to be -1, but that was in error. At some future date, the default will change to -1, as originally intended. Until then, the axis should be given explicitly when arr.ndim > 1, to avoid a FutureWarning.

kind{‘quicksort’, ‘mergesort’, ‘heapsort’, ‘stable’}, optional

The sorting algorithm used.

orderlist, optional

When a is an array with fields defined, this argument specifies which fields to compare first, second, etc. Not all fields need be specified.

endwith{True, False}, optional

Whether missing values (if any) should be treated as the largest values (True) or the smallest values (False) When the array contains unmasked values at the same extremes of the datatype, the ordering of these values and the masked values is undefined.

fill_valuescalar or None, optional

Value used internally for the masked values. If fill_value is not None, it supersedes endwith.

index_arrayndarray, int

Array of indices that sort a along the specified axis. In other words, a[index_array] yields a sorted a.

ma.MaskedArray.sort : Describes sorting algorithms used. lexsort : Indirect stable sort with multiple keys. numpy.ndarray.sort : Inplace sort.

See sort for notes on the different sorting algorithms.

>>> a = np.ma.array([3,2,1], mask=[False, False, True])
>>> a
masked_array(data=[3, 2, --],
             mask=[False, False,  True],
       fill_value=999999)
>>> a.argsort()
array([1, 0, 2])
astype(dtype, order='K', casting='unsafe', subok=True, copy=True)

Copy of the array, cast to a specified type.

dtypestr or dtype

Typecode or data-type to which the array is cast.

order{‘C’, ‘F’, ‘A’, ‘K’}, optional

Controls the memory layout order of the result. ‘C’ means C order, ‘F’ means Fortran order, ‘A’ means ‘F’ order if all the arrays are Fortran contiguous, ‘C’ order otherwise, and ‘K’ means as close to the order the array elements appear in memory as possible. Default is ‘K’.

casting{‘no’, ‘equiv’, ‘safe’, ‘same_kind’, ‘unsafe’}, optional

Controls what kind of data casting may occur. Defaults to ‘unsafe’ for backwards compatibility.

  • ‘no’ means the data types should not be cast at all.

  • ‘equiv’ means only byte-order changes are allowed.

  • ‘safe’ means only casts which can preserve values are allowed.

  • ‘same_kind’ means only safe casts or casts within a kind, like float64 to float32, are allowed.

  • ‘unsafe’ means any data conversions may be done.

subokbool, optional

If True, then sub-classes will be passed-through (default), otherwise the returned array will be forced to be a base-class array.

copybool, optional

By default, astype always returns a newly allocated array. If this is set to false, and the dtype, order, and subok requirements are satisfied, the input array is returned instead of a copy.

arr_tndarray

Unless copy is False and the other conditions for returning the input array are satisfied (see description for copy input parameter), arr_t is a new array of the same shape as the input array, with dtype, order given by dtype, order.

Changed in version 1.17.0: Casting between a simple data type and a structured one is possible only for “unsafe” casting. Casting to multiple fields is allowed, but casting from multiple fields is not.

Changed in version 1.9.0: Casting from numeric to string types in ‘safe’ casting mode requires that the string dtype length is long enough to store the max integer/float value converted.

ComplexWarning

When casting from complex to float or int. To avoid this, one should use a.real.astype(t).

>>> x = np.array([1, 2, 2.5])
>>> x
array([1. ,  2. ,  2.5])
>>> x.astype(int)
array([1, 2, 2])
byteswap(inplace=False)

Swap the bytes of the array elements

Toggle between low-endian and big-endian data representation by returning a byteswapped array, optionally swapped in-place. Arrays of byte-strings are not swapped. The real and imaginary parts of a complex number are swapped individually.

inplacebool, optional

If True, swap bytes in-place, default is False.

outndarray

The byteswapped array. If inplace is True, this is a view to self.

>>> A = np.array([1, 256, 8755], dtype=np.int16)
>>> list(map(hex, A))
['0x1', '0x100', '0x2233']
>>> A.byteswap(inplace=True)
array([  256,     1, 13090], dtype=int16)
>>> list(map(hex, A))
['0x100', '0x1', '0x3322']

Arrays of byte-strings are not swapped

>>> A = np.array([b'ceg', b'fac'])
>>> A.byteswap()
array([b'ceg', b'fac'], dtype='|S3')
A.newbyteorder().byteswap() produces an array with the same values

but different representation in memory

>>> A = np.array([1, 2, 3])
>>> A.view(np.uint8)
array([1, 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 3, 0, 0, 0, 0, 0,
       0, 0], dtype=uint8)
>>> A.newbyteorder().byteswap(inplace=True)
array([1, 2, 3])
>>> A.view(np.uint8)
array([0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0,
       0, 3], dtype=uint8)
choose(choices, out=None, mode='raise')

Use an index array to construct a new array from a set of choices.

Refer to numpy.choose for full documentation.

numpy.choose : equivalent function

clip(min=None, max=None, out=None, **kwargs)

Return an array whose values are limited to [min, max]. One of max or min must be given.

Refer to numpy.clip for full documentation.

numpy.clip : equivalent function

compress(condition, axis=None, out=None)

Return a where condition is True.

If condition is a ~ma.MaskedArray, missing values are considered as False.

conditionvar

Boolean 1-d array selecting which entries to return. If len(condition) is less than the size of a along the axis, then output is truncated to length of condition array.

axis{None, int}, optional

Axis along which the operation must be performed.

out{None, ndarray}, optional

Alternative output array in which to place the result. It must have the same shape as the expected output but the type will be cast if necessary.

resultMaskedArray

A MaskedArray object.

Please note the difference with compressed() ! The output of compress() has a mask, the output of compressed() does not.

>>> x = np.ma.array([[1,2,3],[4,5,6],[7,8,9]], mask=[0] + [1,0]*4)
>>> x
masked_array(
  data=[[1, --, 3],
        [--, 5, --],
        [7, --, 9]],
  mask=[[False,  True, False],
        [ True, False,  True],
        [False,  True, False]],
  fill_value=999999)
>>> x.compress([1, 0, 1])
masked_array(data=[1, 3],
             mask=[False, False],
       fill_value=999999)
>>> x.compress([1, 0, 1], axis=1)
masked_array(
  data=[[1, 3],
        [--, --],
        [7, 9]],
  mask=[[False, False],
        [ True,  True],
        [False, False]],
  fill_value=999999)
compressed()

Return all the non-masked data as a 1-D array.

datandarray

A new ndarray holding the non-masked data is returned.

The result is not a MaskedArray!

>>> x = np.ma.array(np.arange(5), mask=[0]*2 + [1]*3)
>>> x.compressed()
array([0, 1])
>>> type(x.compressed())
<class 'numpy.ndarray'>
conj()

Complex-conjugate all elements.

Refer to numpy.conjugate for full documentation.

numpy.conjugate : equivalent function

conjugate()

Return the complex conjugate, element-wise.

Refer to numpy.conjugate for full documentation.

numpy.conjugate : equivalent function

copy(order='C')

Return a copy of the array.

order{‘C’, ‘F’, ‘A’, ‘K’}, optional

Controls the memory layout of the copy. ‘C’ means C-order, ‘F’ means F-order, ‘A’ means ‘F’ if a is Fortran contiguous, ‘C’ otherwise. ‘K’ means match the layout of a as closely as possible. (Note that this function and numpy.copy() are very similar but have different default values for their order= arguments, and this function always passes sub-classes through.)

numpy.copy : Similar function with different default behavior numpy.copyto

This function is the preferred method for creating an array copy. The function numpy.copy() is similar, but it defaults to using order ‘K’, and will not pass sub-classes through by default.

>>> x = np.array([[1,2,3],[4,5,6]], order='F')
>>> y = x.copy()
>>> x.fill(0)
>>> x
array([[0, 0, 0],
       [0, 0, 0]])
>>> y
array([[1, 2, 3],
       [4, 5, 6]])
>>> y.flags['C_CONTIGUOUS']
True
count(axis=None, keepdims=<no value>)

Count the non-masked elements of the array along the given axis.

axisNone or int or tuple of ints, optional

Axis or axes along which the count is performed. The default, None, performs the count over all the dimensions of the input array. axis may be negative, in which case it counts from the last to the first axis.

New in version 1.10.0.

If this is a tuple of ints, the count is performed on multiple axes, instead of a single axis or all the axes as before.

keepdimsbool, optional

If this is set to True, the axes which are reduced are left in the result as dimensions with size one. With this option, the result will broadcast correctly against the array.

resultndarray or scalar

An array with the same shape as the input array, with the specified axis removed. If the array is a 0-d array, or if axis is None, a scalar is returned.

ma.count_masked : Count masked elements in array or along a given axis.

>>> import numpy.ma as ma
>>> a = ma.arange(6).reshape((2, 3))
>>> a[1, :] = ma.masked
>>> a
masked_array(
  data=[[0, 1, 2],
        [--, --, --]],
  mask=[[False, False, False],
        [ True,  True,  True]],
  fill_value=999999)
>>> a.count()
3

When the axis keyword is specified an array of appropriate size is returned.

>>> a.count(axis=0)
array([1, 1, 1])
>>> a.count(axis=1)
array([3, 0])
cumprod(axis=None, dtype=None, out=None)

Return the cumulative product of the array elements over the given axis.

Masked values are set to 1 internally during the computation. However, their position is saved, and the result will be masked at the same locations.

Refer to numpy.cumprod for full documentation.

The mask is lost if out is not a valid MaskedArray !

Arithmetic is modular when using integer types, and no error is raised on overflow.

numpy.ndarray.cumprod : corresponding function for ndarrays numpy.cumprod : equivalent function

cumsum(axis=None, dtype=None, out=None)

Return the cumulative sum of the array elements over the given axis.

Masked values are set to 0 internally during the computation. However, their position is saved, and the result will be masked at the same locations.

Refer to numpy.cumsum for full documentation.

The mask is lost if out is not a valid ma.MaskedArray !

Arithmetic is modular when using integer types, and no error is raised on overflow.

numpy.ndarray.cumsum : corresponding function for ndarrays numpy.cumsum : equivalent function

>>> marr = np.ma.array(np.arange(10), mask=[0,0,0,1,1,1,0,0,0,0])
>>> marr.cumsum()
masked_array(data=[0, 1, 3, --, --, --, 9, 16, 24, 33],
             mask=[False, False, False,  True,  True,  True, False, False,
                   False, False],
       fill_value=999999)
diagonal(offset=0, axis1=0, axis2=1)

Return specified diagonals. In NumPy 1.9 the returned array is a read-only view instead of a copy as in previous NumPy versions. In a future version the read-only restriction will be removed.

Refer to numpy.diagonal() for full documentation.

numpy.diagonal : equivalent function

dot(b, out=None)

Masked dot product of two arrays. Note that out and strict are located in different positions than in ma.dot. In order to maintain compatibility with the functional version, it is recommended that the optional arguments be treated as keyword only. At some point that may be mandatory.

New in version 1.10.0.

bmasked_array_like

Inputs array.

outmasked_array, optional

Output argument. This must have the exact kind that would be returned if it was not used. In particular, it must have the right type, must be C-contiguous, and its dtype must be the dtype that would be returned for ma.dot(a,b). This is a performance feature. Therefore, if these conditions are not met, an exception is raised, instead of attempting to be flexible.

strictbool, optional

Whether masked data are propagated (True) or set to 0 (False) for the computation. Default is False. Propagating the mask means that if a masked value appears in a row or column, the whole row or column is considered masked.

New in version 1.10.2.

numpy.ma.dot : equivalent function

dump(file)

Dump a pickle of the array to the specified file. The array can be read back with pickle.load or numpy.load.

filestr or Path

A string naming the dump file.

Changed in version 1.17.0: pathlib.Path objects are now accepted.

dumps()

Returns the pickle of the array as a string. pickle.loads or numpy.loads will convert the string back to an array.

None

fill(value)

Fill the array with a scalar value.

valuescalar

All elements of a will be assigned this value.

>>> a = np.array([1, 2])
>>> a.fill(0)
>>> a
array([0, 0])
>>> a = np.empty(2)
>>> a.fill(1)
>>> a
array([1.,  1.])
filled(fill_value=None)

Return a copy of self, with masked values filled with a given value. However, if there are no masked values to fill, self will be returned instead as an ndarray.

fill_valuearray_like, optional

The value to use for invalid entries. Can be scalar or non-scalar. If non-scalar, the resulting ndarray must be broadcastable over input array. Default is None, in which case, the fill_value attribute of the array is used instead.

filled_arrayndarray

A copy of self with invalid entries replaced by fill_value (be it the function argument or the attribute of self), or self itself as an ndarray if there are no invalid entries to be replaced.

The result is not a MaskedArray!

>>> x = np.ma.array([1,2,3,4,5], mask=[0,0,1,0,1], fill_value=-999)
>>> x.filled()
array([   1,    2, -999,    4, -999])
>>> x.filled(fill_value=1000)
array([   1,    2, 1000,    4, 1000])
>>> type(x.filled())
<class 'numpy.ndarray'>

Subclassing is preserved. This means that if, e.g., the data part of the masked array is a recarray, filled returns a recarray:

>>> x = np.array([(-1, 2), (-3, 4)], dtype='i8,i8').view(np.recarray)
>>> m = np.ma.array(x, mask=[(True, False), (False, True)])
>>> m.filled()
rec.array([(999999,      2), (    -3, 999999)],
          dtype=[('f0', '<i8'), ('f1', '<i8')])
flatten(order='C')

Return a copy of the array collapsed into one dimension.

order{‘C’, ‘F’, ‘A’, ‘K’}, optional

‘C’ means to flatten in row-major (C-style) order. ‘F’ means to flatten in column-major (Fortran- style) order. ‘A’ means to flatten in column-major order if a is Fortran contiguous in memory, row-major order otherwise. ‘K’ means to flatten a in the order the elements occur in memory. The default is ‘C’.

yndarray

A copy of the input array, flattened to one dimension.

ravel : Return a flattened array. flat : A 1-D flat iterator over the array.

>>> a = np.array([[1,2], [3,4]])
>>> a.flatten()
array([1, 2, 3, 4])
>>> a.flatten('F')
array([1, 3, 2, 4])
get_fill_value()

The filling value of the masked array is a scalar. When setting, None will set to a default based on the data type.

>>> for dt in [np.int32, np.int64, np.float64, np.complex128]:
...     np.ma.array([0, 1], dtype=dt).get_fill_value()
...
999999
999999
1e+20
(1e+20+0j)
>>> x = np.ma.array([0, 1.], fill_value=-np.inf)
>>> x.fill_value
-inf
>>> x.fill_value = np.pi
>>> x.fill_value
3.1415926535897931 # may vary

Reset to default:

>>> x.fill_value = None
>>> x.fill_value
1e+20
get_imag()

The imaginary part of the masked array.

This property is a view on the imaginary part of this MaskedArray.

real

>>> x = np.ma.array([1+1.j, -2j, 3.45+1.6j], mask=[False, True, False])
>>> x.imag
masked_array(data=[1.0, --, 1.6],
             mask=[False,  True, False],
       fill_value=1e+20)
get_real()

The real part of the masked array.

This property is a view on the real part of this MaskedArray.

imag

>>> x = np.ma.array([1+1.j, -2j, 3.45+1.6j], mask=[False, True, False])
>>> x.real
masked_array(data=[1.0, --, 3.45],
             mask=[False,  True, False],
       fill_value=1e+20)
getfield(dtype, offset=0)

Returns a field of the given array as a certain type.

A field is a view of the array data with a given data-type. The values in the view are determined by the given type and the offset into the current array in bytes. The offset needs to be such that the view dtype fits in the array dtype; for example an array of dtype complex128 has 16-byte elements. If taking a view with a 32-bit integer (4 bytes), the offset needs to be between 0 and 12 bytes.

dtypestr or dtype

The data type of the view. The dtype size of the view can not be larger than that of the array itself.

offsetint

Number of bytes to skip before beginning the element view.

>>> x = np.diag([1.+1.j]*2)
>>> x[1, 1] = 2 + 4.j
>>> x
array([[1.+1.j,  0.+0.j],
       [0.+0.j,  2.+4.j]])
>>> x.getfield(np.float64)
array([[1.,  0.],
       [0.,  2.]])

By choosing an offset of 8 bytes we can select the complex part of the array for our view:

>>> x.getfield(np.float64, offset=8)
array([[1.,  0.],
       [0.,  4.]])
harden_mask()

Force the mask to hard.

Whether the mask of a masked array is hard or soft is determined by its ~ma.MaskedArray.hardmask property. harden_mask sets ~ma.MaskedArray.hardmask to True.

ma.MaskedArray.hardmask

ids()

Return the addresses of the data and mask areas.

None

>>> x = np.ma.array([1, 2, 3], mask=[0, 1, 1])
>>> x.ids()
(166670640, 166659832) # may vary

If the array has no mask, the address of nomask is returned. This address is typically not close to the data in memory:

>>> x = np.ma.array([1, 2, 3])
>>> x.ids()
(166691080, 3083169284) # may vary
iscontiguous()

Return a boolean indicating whether the data is contiguous.

None

>>> x = np.ma.array([1, 2, 3])
>>> x.iscontiguous()
True

iscontiguous returns one of the flags of the masked array:

>>> x.flags
  C_CONTIGUOUS : True
  F_CONTIGUOUS : True
  OWNDATA : False
  WRITEABLE : True
  ALIGNED : True
  WRITEBACKIFCOPY : False
  UPDATEIFCOPY : False
item(*args)

Copy an element of an array to a standard Python scalar and return it.

*args : Arguments (variable number and type)

  • none: in this case, the method only works for arrays with one element (a.size == 1), which element is copied into a standard Python scalar object and returned.

  • int_type: this argument is interpreted as a flat index into the array, specifying which element to copy and return.

  • tuple of int_types: functions as does a single int_type argument, except that the argument is interpreted as an nd-index into the array.

zStandard Python scalar object

A copy of the specified element of the array as a suitable Python scalar

When the data type of a is longdouble or clongdouble, item() returns a scalar array object because there is no available Python scalar that would not lose information. Void arrays return a buffer object for item(), unless fields are defined, in which case a tuple is returned.

item is very similar to a[args], except, instead of an array scalar, a standard Python scalar is returned. This can be useful for speeding up access to elements of the array and doing arithmetic on elements of the array using Python’s optimized math.

>>> np.random.seed(123)
>>> x = np.random.randint(9, size=(3, 3))
>>> x
array([[2, 2, 6],
       [1, 3, 6],
       [1, 0, 1]])
>>> x.item(3)
1
>>> x.item(7)
0
>>> x.item((0, 1))
2
>>> x.item((2, 2))
1
itemset(*args)

Insert scalar into an array (scalar is cast to array’s dtype, if possible)

There must be at least 1 argument, and define the last argument as item. Then, a.itemset(*args) is equivalent to but faster than a[args] = item. The item should be a scalar value and args must select a single item in the array a.

*argsArguments

If one argument: a scalar, only used in case a is of size 1. If two arguments: the last argument is the value to be set and must be a scalar, the first argument specifies a single array element location. It is either an int or a tuple.

Compared to indexing syntax, itemset provides some speed increase for placing a scalar into a particular location in an ndarray, if you must do this. However, generally this is discouraged: among other problems, it complicates the appearance of the code. Also, when using itemset (and item) inside a loop, be sure to assign the methods to a local variable to avoid the attribute look-up at each loop iteration.

>>> np.random.seed(123)
>>> x = np.random.randint(9, size=(3, 3))
>>> x
array([[2, 2, 6],
       [1, 3, 6],
       [1, 0, 1]])
>>> x.itemset(4, 0)
>>> x.itemset((2, 2), 9)
>>> x
array([[2, 2, 6],
       [1, 0, 6],
       [1, 0, 9]])
keys()[source]

Return a list of column headers.

max(axis=None, out=None, fill_value=None, keepdims=<no value>)

Return the maximum along a given axis.

axis{None, int}, optional

Axis along which to operate. By default, axis is None and the flattened input is used.

outarray_like, optional

Alternative output array in which to place the result. Must be of the same shape and buffer length as the expected output.

fill_valuescalar or None, optional

Value used to fill in the masked values. If None, use the output of maximum_fill_value().

keepdimsbool, optional

If this is set to True, the axes which are reduced are left in the result as dimensions with size one. With this option, the result will broadcast correctly against the array.

amaxarray_like

New array holding the result. If out was specified, out is returned.

ma.maximum_fill_value

Returns the maximum filling value for a given datatype.

mean(axis=None, dtype=None, out=None, keepdims=<no value>)

Returns the average of the array elements along given axis.

Masked entries are ignored, and result elements which are not finite will be masked.

Refer to numpy.mean for full documentation.

numpy.ndarray.mean : corresponding function for ndarrays numpy.mean : Equivalent function numpy.ma.average : Weighted average.

>>> a = np.ma.array([1,2,3], mask=[False, False, True])
>>> a
masked_array(data=[1, 2, --],
             mask=[False, False,  True],
       fill_value=999999)
>>> a.mean()
1.5
min(axis=None, out=None, fill_value=None, keepdims=<no value>)

Return the minimum along a given axis.

axis{None, int}, optional

Axis along which to operate. By default, axis is None and the flattened input is used.

outarray_like, optional

Alternative output array in which to place the result. Must be of the same shape and buffer length as the expected output.

fill_valuescalar or None, optional

Value used to fill in the masked values. If None, use the output of minimum_fill_value.

keepdimsbool, optional

If this is set to True, the axes which are reduced are left in the result as dimensions with size one. With this option, the result will broadcast correctly against the array.

aminarray_like

New array holding the result. If out was specified, out is returned.

ma.minimum_fill_value

Returns the minimum filling value for a given datatype.

mini(axis=None)

Return the array minimum along the specified axis.

Deprecated since version 1.13.0: This function is identical to both:

  • self.min(keepdims=True, axis=axis).squeeze(axis=axis)

  • np.ma.minimum.reduce(self, axis=axis)

Typically though, self.min(axis=axis) is sufficient.

axisint, optional

The axis along which to find the minima. Default is None, in which case the minimum value in the whole array is returned.

minscalar or MaskedArray

If axis is None, the result is a scalar. Otherwise, if axis is given and the array is at least 2-D, the result is a masked array with dimension one smaller than the array on which mini is called.

>>> x = np.ma.array(np.arange(6), mask=[0 ,1, 0, 0, 0 ,1]).reshape(3, 2)
>>> x
masked_array(
  data=[[0, --],
        [2, 3],
        [4, --]],
  mask=[[False,  True],
        [False, False],
        [False,  True]],
  fill_value=999999)
>>> x.mini()
masked_array(data=0,
             mask=False,
       fill_value=999999)
>>> x.mini(axis=0)
masked_array(data=[0, 3],
             mask=[False, False],
       fill_value=999999)
>>> x.mini(axis=1)
masked_array(data=[0, 2, 4],
             mask=[False, False, False],
       fill_value=999999)

There is a small difference between mini and min:

>>> x[:,1].mini(axis=0)
masked_array(data=3,
             mask=False,
       fill_value=999999)
>>> x[:,1].min(axis=0)
3
newbyteorder(new_order='S', /)

Return the array with the same data viewed with a different byte order.

Equivalent to:

arr.view(arr.dtype.newbytorder(new_order))

Changes are also made in all fields and sub-arrays of the array data type.

new_orderstring, optional

Byte order to force; a value from the byte order specifications below. new_order codes can be any of:

  • ‘S’ - swap dtype from current to opposite endian

  • {‘<’, ‘little’} - little endian

  • {‘>’, ‘big’} - big endian

  • ‘=’ - native order, equivalent to sys.byteorder

  • {‘|’, ‘I’} - ignore (no change to byte order)

The default value (‘S’) results in swapping the current byte order.

new_arrarray

New array object with the dtype reflecting given change to the byte order.

nonzero()

Return the indices of unmasked elements that are not zero.

Returns a tuple of arrays, one for each dimension, containing the indices of the non-zero elements in that dimension. The corresponding non-zero values can be obtained with:

a[a.nonzero()]

To group the indices by element, rather than dimension, use instead:

np.transpose(a.nonzero())

The result of this is always a 2d array, with a row for each non-zero element.

None

tuple_of_arraystuple

Indices of elements that are non-zero.

numpy.nonzero :

Function operating on ndarrays.

flatnonzero :

Return indices that are non-zero in the flattened version of the input array.

numpy.ndarray.nonzero :

Equivalent ndarray method.

count_nonzero :

Counts the number of non-zero elements in the input array.

>>> import numpy.ma as ma
>>> x = ma.array(np.eye(3))
>>> x
masked_array(
  data=[[1., 0., 0.],
        [0., 1., 0.],
        [0., 0., 1.]],
  mask=False,
  fill_value=1e+20)
>>> x.nonzero()
(array([0, 1, 2]), array([0, 1, 2]))

Masked elements are ignored.

>>> x[1, 1] = ma.masked
>>> x
masked_array(
  data=[[1.0, 0.0, 0.0],
        [0.0, --, 0.0],
        [0.0, 0.0, 1.0]],
  mask=[[False, False, False],
        [False,  True, False],
        [False, False, False]],
  fill_value=1e+20)
>>> x.nonzero()
(array([0, 2]), array([0, 2]))

Indices can also be grouped by element.

>>> np.transpose(x.nonzero())
array([[0, 0],
       [2, 2]])

A common use for nonzero is to find the indices of an array, where a condition is True. Given an array a, the condition a > 3 is a boolean array and since False is interpreted as 0, ma.nonzero(a > 3) yields the indices of the a where the condition is true.

>>> a = ma.array([[1,2,3],[4,5,6],[7,8,9]])
>>> a > 3
masked_array(
  data=[[False, False, False],
        [ True,  True,  True],
        [ True,  True,  True]],
  mask=False,
  fill_value=True)
>>> ma.nonzero(a > 3)
(array([1, 1, 1, 2, 2, 2]), array([0, 1, 2, 0, 1, 2]))

The nonzero method of the condition array can also be called.

>>> (a > 3).nonzero()
(array([1, 1, 1, 2, 2, 2]), array([0, 1, 2, 0, 1, 2]))
partition(kth, axis=- 1, kind='introselect', order=None)

Rearranges the elements in the array in such a way that the value of the element in kth position is in the position it would be in a sorted array. All elements smaller than the kth element are moved before this element and all equal or greater are moved behind it. The ordering of the elements in the two partitions is undefined.

New in version 1.8.0.

kthint or sequence of ints

Element index to partition by. The kth element value will be in its final sorted position and all smaller elements will be moved before it and all equal or greater elements behind it. The order of all elements in the partitions is undefined. If provided with a sequence of kth it will partition all elements indexed by kth of them into their sorted position at once.

axisint, optional

Axis along which to sort. Default is -1, which means sort along the last axis.

kind{‘introselect’}, optional

Selection algorithm. Default is ‘introselect’.

orderstr or list of str, optional

When a is an array with fields defined, this argument specifies which fields to compare first, second, etc. A single field can be specified as a string, and not all fields need to be specified, but unspecified fields will still be used, in the order in which they come up in the dtype, to break ties.

numpy.partition : Return a parititioned copy of an array. argpartition : Indirect partition. sort : Full sort.

See np.partition for notes on the different algorithms.

>>> a = np.array([3, 4, 2, 1])
>>> a.partition(3)
>>> a
array([2, 1, 3, 4])
>>> a.partition((1, 3))
>>> a
array([1, 2, 3, 4])
prod(axis=None, dtype=None, out=None, keepdims=<no value>)

Return the product of the array elements over the given axis.

Masked elements are set to 1 internally for computation.

Refer to numpy.prod for full documentation.

Arithmetic is modular when using integer types, and no error is raised on overflow.

numpy.ndarray.prod : corresponding function for ndarrays numpy.prod : equivalent function

product(axis=None, dtype=None, out=None, keepdims=<no value>)

Return the product of the array elements over the given axis.

Masked elements are set to 1 internally for computation.

Refer to numpy.prod for full documentation.

Arithmetic is modular when using integer types, and no error is raised on overflow.

numpy.ndarray.prod : corresponding function for ndarrays numpy.prod : equivalent function

ptp(axis=None, out=None, fill_value=None, keepdims=False)

Return (maximum - minimum) along the given dimension (i.e. peak-to-peak value).

Warning

ptp preserves the data type of the array. This means the return value for an input of signed integers with n bits (e.g. np.int8, np.int16, etc) is also a signed integer with n bits. In that case, peak-to-peak values greater than 2**(n-1)-1 will be returned as negative values. An example with a work-around is shown below.

axis{None, int}, optional

Axis along which to find the peaks. If None (default) the flattened array is used.

out{None, array_like}, optional

Alternative output array in which to place the result. It must have the same shape and buffer length as the expected output but the type will be cast if necessary.

fill_valuescalar or None, optional

Value used to fill in the masked values.

keepdimsbool, optional

If this is set to True, the axes which are reduced are left in the result as dimensions with size one. With this option, the result will broadcast correctly against the array.

ptpndarray.

A new array holding the result, unless out was specified, in which case a reference to out is returned.

>>> x = np.ma.MaskedArray([[4, 9, 2, 10],
...                        [6, 9, 7, 12]])
>>> x.ptp(axis=1)
masked_array(data=[8, 6],
             mask=False,
       fill_value=999999)
>>> x.ptp(axis=0)
masked_array(data=[2, 0, 5, 2],
             mask=False,
       fill_value=999999)
>>> x.ptp()
10

This example shows that a negative value can be returned when the input is an array of signed integers.

>>> y = np.ma.MaskedArray([[1, 127],
...                        [0, 127],
...                        [-1, 127],
...                        [-2, 127]], dtype=np.int8)
>>> y.ptp(axis=1)
masked_array(data=[ 126,  127, -128, -127],
             mask=False,
       fill_value=999999,
            dtype=int8)

A work-around is to use the view() method to view the result as unsigned integers with the same bit width:

>>> y.ptp(axis=1).view(np.uint8)
masked_array(data=[126, 127, 128, 129],
             mask=False,
       fill_value=999999,
            dtype=uint8)
put(indices, values, mode='raise')

Set storage-indexed locations to corresponding values.

Sets self._data.flat[n] = values[n] for each n in indices. If values is shorter than indices then it will repeat. If values has some masked values, the initial mask is updated in consequence, else the corresponding values are unmasked.

indices1-D array_like

Target indices, interpreted as integers.

valuesarray_like

Values to place in self._data copy at target indices.

mode{‘raise’, ‘wrap’, ‘clip’}, optional

Specifies how out-of-bounds indices will behave. ‘raise’ : raise an error. ‘wrap’ : wrap around. ‘clip’ : clip to the range.

values can be a scalar or length 1 array.

>>> x = np.ma.array([[1,2,3],[4,5,6],[7,8,9]], mask=[0] + [1,0]*4)
>>> x
masked_array(
  data=[[1, --, 3],
        [--, 5, --],
        [7, --, 9]],
  mask=[[False,  True, False],
        [ True, False,  True],
        [False,  True, False]],
  fill_value=999999)
>>> x.put([0,4,8],[10,20,30])
>>> x
masked_array(
  data=[[10, --, 3],
        [--, 20, --],
        [7, --, 30]],
  mask=[[False,  True, False],
        [ True, False,  True],
        [False,  True, False]],
  fill_value=999999)
>>> x.put(4,999)
>>> x
masked_array(
  data=[[10, --, 3],
        [--, 999, --],
        [7, --, 30]],
  mask=[[False,  True, False],
        [ True, False,  True],
        [False,  True, False]],
  fill_value=999999)
ravel(order='C')

Returns a 1D version of self, as a view.

order{‘C’, ‘F’, ‘A’, ‘K’}, optional

The elements of a are read using this index order. ‘C’ means to index the elements in C-like order, with the last axis index changing fastest, back to the first axis index changing slowest. ‘F’ means to index the elements in Fortran-like index order, with the first index changing fastest, and the last index changing slowest. Note that the ‘C’ and ‘F’ options take no account of the memory layout of the underlying array, and only refer to the order of axis indexing. ‘A’ means to read the elements in Fortran-like index order if m is Fortran contiguous in memory, C-like order otherwise. ‘K’ means to read the elements in the order they occur in memory, except for reversing the data when strides are negative. By default, ‘C’ index order is used.

MaskedArray

Output view is of shape (self.size,) (or (np.ma.product(self.shape),)).

>>> x = np.ma.array([[1,2,3],[4,5,6],[7,8,9]], mask=[0] + [1,0]*4)
>>> x
masked_array(
  data=[[1, --, 3],
        [--, 5, --],
        [7, --, 9]],
  mask=[[False,  True, False],
        [ True, False,  True],
        [False,  True, False]],
  fill_value=999999)
>>> x.ravel()
masked_array(data=[1, --, 3, --, 5, --, 7, --, 9],
             mask=[False,  True, False,  True, False,  True, False,  True,
                   False],
       fill_value=999999)
repeat(repeats, axis=None)

Repeat elements of an array.

Refer to numpy.repeat for full documentation.

numpy.repeat : equivalent function

reshape(*s, **kwargs)

Give a new shape to the array without changing its data.

Returns a masked array containing the same data, but with a new shape. The result is a view on the original array; if this is not possible, a ValueError is raised.

shapeint or tuple of ints

The new shape should be compatible with the original shape. If an integer is supplied, then the result will be a 1-D array of that length.

order{‘C’, ‘F’}, optional

Determines whether the array data should be viewed as in C (row-major) or FORTRAN (column-major) order.

reshaped_arrayarray

A new view on the array.

reshape : Equivalent function in the masked array module. numpy.ndarray.reshape : Equivalent method on ndarray object. numpy.reshape : Equivalent function in the NumPy module.

The reshaping operation cannot guarantee that a copy will not be made, to modify the shape in place, use a.shape = s

>>> x = np.ma.array([[1,2],[3,4]], mask=[1,0,0,1])
>>> x
masked_array(
  data=[[--, 2],
        [3, --]],
  mask=[[ True, False],
        [False,  True]],
  fill_value=999999)
>>> x = x.reshape((4,1))
>>> x
masked_array(
  data=[[--],
        [2],
        [3],
        [--]],
  mask=[[ True],
        [False],
        [False],
        [ True]],
  fill_value=999999)
resize(newshape, refcheck=True, order=False)

Warning

This method does nothing, except raise a ValueError exception. A masked array does not own its data and therefore cannot safely be resized in place. Use the numpy.ma.resize function instead.

This method is difficult to implement safely and may be deprecated in future releases of NumPy.

round(decimals=0, out=None)

Return each element rounded to the given number of decimals.

Refer to numpy.around for full documentation.

numpy.ndarray.round : corresponding function for ndarrays numpy.around : equivalent function

searchsorted(v, side='left', sorter=None)

Find indices where elements of v should be inserted in a to maintain order.

For full documentation, see numpy.searchsorted

numpy.searchsorted : equivalent function

set_fill_value(value=None)
setfield(val, dtype, offset=0)

Put a value into a specified place in a field defined by a data-type.

Place val into a’s field defined by dtype and beginning offset bytes into the field.

valobject

Value to be placed in field.

dtypedtype object

Data-type of the field in which to place val.

offsetint, optional

The number of bytes into the field at which to place val.

None

getfield

>>> x = np.eye(3)
>>> x.getfield(np.float64)
array([[1.,  0.,  0.],
       [0.,  1.,  0.],
       [0.,  0.,  1.]])
>>> x.setfield(3, np.int32)
>>> x.getfield(np.int32)
array([[3, 3, 3],
       [3, 3, 3],
       [3, 3, 3]], dtype=int32)
>>> x
array([[1.0e+000, 1.5e-323, 1.5e-323],
       [1.5e-323, 1.0e+000, 1.5e-323],
       [1.5e-323, 1.5e-323, 1.0e+000]])
>>> x.setfield(np.eye(3), np.int32)
>>> x
array([[1.,  0.,  0.],
       [0.,  1.,  0.],
       [0.,  0.,  1.]])
setflags(write=None, align=None, uic=None)

Set array flags WRITEABLE, ALIGNED, (WRITEBACKIFCOPY and UPDATEIFCOPY), respectively.

These Boolean-valued flags affect how numpy interprets the memory area used by a (see Notes below). The ALIGNED flag can only be set to True if the data is actually aligned according to the type. The WRITEBACKIFCOPY and (deprecated) UPDATEIFCOPY flags can never be set to True. The flag WRITEABLE can only be set to True if the array owns its own memory, or the ultimate owner of the memory exposes a writeable buffer interface, or is a string. (The exception for string is made so that unpickling can be done without copying memory.)

writebool, optional

Describes whether or not a can be written to.

alignbool, optional

Describes whether or not a is aligned properly for its type.

uicbool, optional

Describes whether or not a is a copy of another “base” array.

Array flags provide information about how the memory area used for the array is to be interpreted. There are 7 Boolean flags in use, only four of which can be changed by the user: WRITEBACKIFCOPY, UPDATEIFCOPY, WRITEABLE, and ALIGNED.

WRITEABLE (W) the data area can be written to;

ALIGNED (A) the data and strides are aligned appropriately for the hardware (as determined by the compiler);

UPDATEIFCOPY (U) (deprecated), replaced by WRITEBACKIFCOPY;

WRITEBACKIFCOPY (X) this array is a copy of some other array (referenced by .base). When the C-API function PyArray_ResolveWritebackIfCopy is called, the base array will be updated with the contents of this array.

All flags can be accessed using the single (upper case) letter as well as the full name.

>>> y = np.array([[3, 1, 7],
...               [2, 0, 0],
...               [8, 5, 9]])
>>> y
array([[3, 1, 7],
       [2, 0, 0],
       [8, 5, 9]])
>>> y.flags
  C_CONTIGUOUS : True
  F_CONTIGUOUS : False
  OWNDATA : True
  WRITEABLE : True
  ALIGNED : True
  WRITEBACKIFCOPY : False
  UPDATEIFCOPY : False
>>> y.setflags(write=0, align=0)
>>> y.flags
  C_CONTIGUOUS : True
  F_CONTIGUOUS : False
  OWNDATA : True
  WRITEABLE : False
  ALIGNED : False
  WRITEBACKIFCOPY : False
  UPDATEIFCOPY : False
>>> y.setflags(uic=1)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: cannot set WRITEBACKIFCOPY flag to True
shrink_mask()

Reduce a mask to nomask when possible.

None

None

>>> x = np.ma.array([[1,2 ], [3, 4]], mask=[0]*4)
>>> x.mask
array([[False, False],
       [False, False]])
>>> x.shrink_mask()
masked_array(
  data=[[1, 2],
        [3, 4]],
  mask=False,
  fill_value=999999)
>>> x.mask
False
soften_mask()

Force the mask to soft.

Whether the mask of a masked array is hard or soft is determined by its ~ma.MaskedArray.hardmask property. soften_mask sets ~ma.MaskedArray.hardmask to False.

ma.MaskedArray.hardmask

sort(axis=- 1, kind=None, order=None, endwith=True, fill_value=None)

Sort the array, in-place

aarray_like

Array to be sorted.

axisint, optional

Axis along which to sort. If None, the array is flattened before sorting. The default is -1, which sorts along the last axis.

kind{‘quicksort’, ‘mergesort’, ‘heapsort’, ‘stable’}, optional

The sorting algorithm used.

orderlist, optional

When a is a structured array, this argument specifies which fields to compare first, second, and so on. This list does not need to include all of the fields.

endwith{True, False}, optional

Whether missing values (if any) should be treated as the largest values (True) or the smallest values (False) When the array contains unmasked values sorting at the same extremes of the datatype, the ordering of these values and the masked values is undefined.

fill_valuescalar or None, optional

Value used internally for the masked values. If fill_value is not None, it supersedes endwith.

sorted_arrayndarray

Array of the same type and shape as a.

numpy.ndarray.sort : Method to sort an array in-place. argsort : Indirect sort. lexsort : Indirect stable sort on multiple keys. searchsorted : Find elements in a sorted array.

See sort for notes on the different sorting algorithms.

>>> a = np.ma.array([1, 2, 5, 4, 3],mask=[0, 1, 0, 1, 0])
>>> # Default
>>> a.sort()
>>> a
masked_array(data=[1, 3, 5, --, --],
             mask=[False, False, False,  True,  True],
       fill_value=999999)
>>> a = np.ma.array([1, 2, 5, 4, 3],mask=[0, 1, 0, 1, 0])
>>> # Put missing values in the front
>>> a.sort(endwith=False)
>>> a
masked_array(data=[--, --, 1, 3, 5],
             mask=[ True,  True, False, False, False],
       fill_value=999999)
>>> a = np.ma.array([1, 2, 5, 4, 3],mask=[0, 1, 0, 1, 0])
>>> # fill_value takes over endwith
>>> a.sort(endwith=False, fill_value=3)
>>> a
masked_array(data=[1, --, --, 3, 5],
             mask=[False,  True,  True, False, False],
       fill_value=999999)
squeeze(axis=None)

Remove axes of length one from a.

Refer to numpy.squeeze for full documentation.

numpy.squeeze : equivalent function

std(axis=None, dtype=None, out=None, ddof=0, keepdims=<no value>)

Returns the standard deviation of the array elements along given axis.

Masked entries are ignored.

Refer to numpy.std for full documentation.

numpy.ndarray.std : corresponding function for ndarrays numpy.std : Equivalent function

sum(axis=None, dtype=None, out=None, keepdims=<no value>)

Return the sum of the array elements over the given axis.

Masked elements are set to 0 internally.

Refer to numpy.sum for full documentation.

numpy.ndarray.sum : corresponding function for ndarrays numpy.sum : equivalent function

>>> x = np.ma.array([[1,2,3],[4,5,6],[7,8,9]], mask=[0] + [1,0]*4)
>>> x
masked_array(
  data=[[1, --, 3],
        [--, 5, --],
        [7, --, 9]],
  mask=[[False,  True, False],
        [ True, False,  True],
        [False,  True, False]],
  fill_value=999999)
>>> x.sum()
25
>>> x.sum(axis=1)
masked_array(data=[4, 5, 16],
             mask=[False, False, False],
       fill_value=999999)
>>> x.sum(axis=0)
masked_array(data=[8, 5, 12],
             mask=[False, False, False],
       fill_value=999999)
>>> print(type(x.sum(axis=0, dtype=np.int64)[0]))
<class 'numpy.int64'>
swap_column(*swp, **kargs)[source]

Swap pairs of columns in the data.

Useful for reordering data for idiot programs that expect columns in a fixed order.

Parameters
  • swp (tuple of list of tuples of two elements) – Each element will be iused as a column index (using the normal rules for matching columns). The two elements represent the two columns that are to be swapped.

  • headers_too (bool) – Indicates the column headers are swapped as well

Returns

self – A copy of the modified DataFile objects

Note

If swp is a list, then the function is called recursively on each element of the list. Thus in principle the @swp could contain lists of lists of tuples

swapaxes(axis1, axis2)

Return a view of the array with axis1 and axis2 interchanged.

Refer to numpy.swapaxes for full documentation.

numpy.swapaxes : equivalent function

take(indices, axis=None, out=None, mode='raise')
tobytes(fill_value=None, order='C')

Return the array data as a string containing the raw bytes in the array.

The array is filled with a fill value before the string conversion.

New in version 1.9.0.

fill_valuescalar, optional

Value used to fill in the masked values. Default is None, in which case MaskedArray.fill_value is used.

order{‘C’,’F’,’A’}, optional

Order of the data item in the copy. Default is ‘C’.

  • ‘C’ – C order (row major).

  • ‘F’ – Fortran order (column major).

  • ‘A’ – Any, current order of array.

  • None – Same as ‘A’.

numpy.ndarray.tobytes tolist, tofile

As for ndarray.tobytes, information about the shape, dtype, etc., but also about fill_value, will be lost.

>>> x = np.ma.array(np.array([[1, 2], [3, 4]]), mask=[[0, 1], [1, 0]])
>>> x.tobytes()
b'\x01\x00\x00\x00\x00\x00\x00\x00?B\x0f\x00\x00\x00\x00\x00?B\x0f\x00\x00\x00\x00\x00\x04\x00\x00\x00\x00\x00\x00\x00'
tofile(fid, sep='', format='%s')

Save a masked array to a file in binary format.

Warning

This function is not implemented yet.

NotImplementedError

When tofile is called.

toflex()

Transforms a masked array into a flexible-type array.

The flexible type array that is returned will have two fields:

  • the _data field stores the _data part of the array.

  • the _mask field stores the _mask part of the array.

None

recordndarray

A new flexible-type ndarray with two fields: the first element containing a value, the second element containing the corresponding mask boolean. The returned record shape matches self.shape.

A side-effect of transforming a masked array into a flexible ndarray is that meta information (fill_value, …) will be lost.

>>> x = np.ma.array([[1,2,3],[4,5,6],[7,8,9]], mask=[0] + [1,0]*4)
>>> x
masked_array(
  data=[[1, --, 3],
        [--, 5, --],
        [7, --, 9]],
  mask=[[False,  True, False],
        [ True, False,  True],
        [False,  True, False]],
  fill_value=999999)
>>> x.toflex()
array([[(1, False), (2,  True), (3, False)],
       [(4,  True), (5, False), (6,  True)],
       [(7, False), (8,  True), (9, False)]],
      dtype=[('_data', '<i8'), ('_mask', '?')])
tolist(fill_value=None)

Return the data portion of the masked array as a hierarchical Python list.

Data items are converted to the nearest compatible Python type. Masked values are converted to fill_value. If fill_value is None, the corresponding entries in the output list will be None.

fill_valuescalar, optional

The value to use for invalid entries. Default is None.

resultlist

The Python list representation of the masked array.

>>> x = np.ma.array([[1,2,3], [4,5,6], [7,8,9]], mask=[0] + [1,0]*4)
>>> x.tolist()
[[1, None, 3], [None, 5, None], [7, None, 9]]
>>> x.tolist(-999)
[[1, -999, 3], [-999, 5, -999], [7, -999, 9]]
torecords()

Transforms a masked array into a flexible-type array.

The flexible type array that is returned will have two fields:

  • the _data field stores the _data part of the array.

  • the _mask field stores the _mask part of the array.

None

recordndarray

A new flexible-type ndarray with two fields: the first element containing a value, the second element containing the corresponding mask boolean. The returned record shape matches self.shape.

A side-effect of transforming a masked array into a flexible ndarray is that meta information (fill_value, …) will be lost.

>>> x = np.ma.array([[1,2,3],[4,5,6],[7,8,9]], mask=[0] + [1,0]*4)
>>> x
masked_array(
  data=[[1, --, 3],
        [--, 5, --],
        [7, --, 9]],
  mask=[[False,  True, False],
        [ True, False,  True],
        [False,  True, False]],
  fill_value=999999)
>>> x.toflex()
array([[(1, False), (2,  True), (3, False)],
       [(4,  True), (5, False), (6,  True)],
       [(7, False), (8,  True), (9, False)]],
      dtype=[('_data', '<i8'), ('_mask', '?')])
tostring(fill_value=None, order='C')

A compatibility alias for tobytes, with exactly the same behavior.

Despite its name, it returns bytes not strs.

Deprecated since version 1.19.0.

trace(offset=0, axis1=0, axis2=1, dtype=None, out=None)

Return the sum along diagonals of the array.

Refer to numpy.trace for full documentation.

numpy.trace : equivalent function

transpose(*axes)

Returns a view of the array with axes transposed.

For a 1-D array this has no effect, as a transposed vector is simply the same vector. To convert a 1-D array into a 2D column vector, an additional dimension must be added. np.atleast2d(a).T achieves this, as does a[:, np.newaxis]. For a 2-D array, this is a standard matrix transpose. For an n-D array, if axes are given, their order indicates how the axes are permuted (see Examples). If axes are not provided and a.shape = (i[0], i[1], ... i[n-2], i[n-1]), then a.transpose().shape = (i[n-1], i[n-2], ... i[1], i[0]).

axes : None, tuple of ints, or n ints

  • None or no argument: reverses the order of the axes.

  • tuple of ints: i in the j-th place in the tuple means a’s i-th axis becomes a.transpose()’s j-th axis.

  • n ints: same as an n-tuple of the same ints (this form is intended simply as a “convenience” alternative to the tuple form)

outndarray

View of a, with axes suitably permuted.

transpose : Equivalent function ndarray.T : Array property returning the array transposed. ndarray.reshape : Give a new shape to an array without changing its data.

>>> a = np.array([[1, 2], [3, 4]])
>>> a
array([[1, 2],
       [3, 4]])
>>> a.transpose()
array([[1, 3],
       [2, 4]])
>>> a.transpose((1, 0))
array([[1, 3],
       [2, 4]])
>>> a.transpose(1, 0)
array([[1, 3],
       [2, 4]])
unshare_mask()

Copy the mask and set the sharedmask flag to False.

Whether the mask is shared between masked arrays can be seen from the sharedmask property. unshare_mask ensures the mask is not shared. A copy of the mask is only made if it was shared.

sharedmask

var(axis=None, dtype=None, out=None, ddof=0, keepdims=<no value>)

Compute the variance along the specified axis.

Returns the variance of the array elements, a measure of the spread of a distribution. The variance is computed for the flattened array by default, otherwise over the specified axis.

aarray_like

Array containing numbers whose variance is desired. If a is not an array, a conversion is attempted.

axisNone or int or tuple of ints, optional

Axis or axes along which the variance is computed. The default is to compute the variance of the flattened array.

New in version 1.7.0.

If this is a tuple of ints, a variance is performed over multiple axes, instead of a single axis or all the axes as before.

dtypedata-type, optional

Type to use in computing the variance. For arrays of integer type the default is float64; for arrays of float types it is the same as the array type.

outndarray, optional

Alternate output array in which to place the result. It must have the same shape as the expected output, but the type is cast if necessary.

ddofint, optional

“Delta Degrees of Freedom”: the divisor used in the calculation is N - ddof, where N represents the number of elements. By default ddof is zero.

keepdimsbool, optional

If this is set to True, the axes which are reduced are left in the result as dimensions with size one. With this option, the result will broadcast correctly against the input array.

If the default value is passed, then keepdims will not be passed through to the var method of sub-classes of ndarray, however any non-default value will be. If the sub-class’ method does not implement keepdims any exceptions will be raised.

wherearray_like of bool, optional

Elements to include in the variance. See ~numpy.ufunc.reduce for details.

New in version 1.20.0.

variancendarray, see dtype parameter above

If out=None, returns a new array containing the variance; otherwise, a reference to the output array is returned.

std, mean, nanmean, nanstd, nanvar Output type determination

The variance is the average of the squared deviations from the mean, i.e., var = mean(x), where x = abs(a - a.mean())**2.

The mean is typically calculated as x.sum() / N, where N = len(x). If, however, ddof is specified, the divisor N - ddof is used instead. In standard statistical practice, ddof=1 provides an unbiased estimator of the variance of a hypothetical infinite population. ddof=0 provides a maximum likelihood estimate of the variance for normally distributed variables.

Note that for complex numbers, the absolute value is taken before squaring, so that the result is always real and nonnegative.

For floating-point input, the variance is computed using the same precision the input has. Depending on the input data, this can cause the results to be inaccurate, especially for float32 (see example below). Specifying a higher-accuracy accumulator using the dtype keyword can alleviate this issue.

>>> a = np.array([[1, 2], [3, 4]])
>>> np.var(a)
1.25
>>> np.var(a, axis=0)
array([1.,  1.])
>>> np.var(a, axis=1)
array([0.25,  0.25])

In single precision, var() can be inaccurate:

>>> a = np.zeros((2, 512*512), dtype=np.float32)
>>> a[0, :] = 1.0
>>> a[1, :] = 0.1
>>> np.var(a)
0.20250003

Computing the variance in float64 is more accurate:

>>> np.var(a, dtype=np.float64)
0.20249999932944759 # may vary
>>> ((1-0.55)**2 + (0.1-0.55)**2)/2
0.2025

Specifying a where argument:

>>> a = np.array([[14, 8, 11, 10], [7, 9, 10, 11], [10, 15, 5, 10]])
>>> np.var(a)
6.833333333333333 # may vary
>>> np.var(a, where=[[True], [True], [False]])
4.0
view(dtype=None, type=None, fill_value=None)

Return a view of the MaskedArray data.

dtypedata-type or ndarray sub-class, optional

Data-type descriptor of the returned view, e.g., float32 or int16. The default, None, results in the view having the same data-type as a. As with ndarray.view, dtype can also be specified as an ndarray sub-class, which then specifies the type of the returned object (this is equivalent to setting the type parameter).

typePython type, optional

Type of the returned view, either ndarray or a subclass. The default None results in type preservation.

fill_valuescalar, optional

The value to use for invalid entries (None by default). If None, then this argument is inferred from the passed dtype, or in its absence the original array, as discussed in the notes below.

numpy.ndarray.view : Equivalent method on ndarray object.

a.view() is used two different ways:

a.view(some_dtype) or a.view(dtype=some_dtype) constructs a view of the array’s memory with a different data-type. This can cause a reinterpretation of the bytes of memory.

a.view(ndarray_subclass) or a.view(type=ndarray_subclass) just returns an instance of ndarray_subclass that looks at the same array (same shape, dtype, etc.) This does not cause a reinterpretation of the memory.

If fill_value is not specified, but dtype is specified (and is not an ndarray sub-class), the fill_value of the MaskedArray will be reset. If neither fill_value nor dtype are specified (or if dtype is an ndarray sub-class), then the fill value is preserved. Finally, if fill_value is specified, but dtype is not, the fill value is set to the specified value.

For a.view(some_dtype), if some_dtype has a different number of bytes per entry than the previous dtype (for example, converting a regular array to a structured array), then the behavior of the view cannot be predicted just from the superficial appearance of a (shown by print(a)). It also depends on exactly how a is stored in memory. Therefore if a is C-ordered versus fortran-ordered, versus defined as a slice or transpose, etc., the view may give different results.