exa.core.container module

Container

The Container class is the primary object for data processing, analysis, and visualization. In brief, containers are composed of data objects whose contents are used for 2D and 3D visualization. Containers also provide some content management and data relationship features.

See also

For a description of data objects see numerical.

class exa.core.container.Container(name=None, description=None, meta=None, uuid=None, **kwargs)

Bases: object

Container class responsible for all features related to data management.

property log

copy(name=None, description=None, meta=None)

Create a copy of the current object (may alter the container’s name, description, and update the metadata if needed).

concat(*args, **kwargs)

Concatenate any number of container objects with the current object into a single container object.

See also

For argument description, see concat().

slice_naive(key)

Naively slice each data object in the container by the object’s index.

Parameters

key – Int, slice, or list by which to extra “sub”-container

Returns

sub – Sub container of the same format with a view of the data

Warning

To ensure that a new container is created, use the copy method.

mycontainer[slice].copy()

slice_cardinal(key)

Slice the container according to its (primary) cardinal axis.

The “cardinal” axis can have any name so long as the name matches a data object attached to the container. The index name for this object should also match the value of the cardinal axis.

The algorithm builds a network graph representing the data relationships (including information about the type of relationship) and then traverses the edge tree (starting from the cardinal table). Each subsequent child object in the tree is sliced based on its relationship with its parent.

Note

Breadth first traversal is performed.

Warning

This function does not make a copy (if possible): to ensure a new object is created (a copy) use copy() after slicing.

myslice = mycontainer[::2].copy()

See also

For data network generation, see network(). For information about relationships between data objects see numerical.

cardinal_groupby()

Create an instance of this class for every step in the cardinal dimension.

info()

Display information about the container’s data objects (note that info on metadata and visualization objects is also provided).

Note

Sizes are reported in bytes.

memory_usage(string=False)

Get the memory usage estimate of the container.

Parameters

string (bool) – Human readable string (default false)

See also

info()

network(figsize=(14, 9), fig=True)

Display information about the container’s object relationships.

Nodes correspond to data objects. The size of the node corresponds to the size of the table in memory. The color of the node corresponds to its fundamental data type. Nodes are labeled by their container name; class information is listed below. The color of the connections correspond to the type of relationship; either an index of one table corresponds to a column in another table or the two tables share an index.

Parameters

• figsize (tuple) – Tuple containing figure dimensions

• fig (bool) – Generate the figure (default true)

Returns

graph – Network graph object containing data relationships

save(path=None, complevel=1, complib='zlib')

Save the container as an HDF5 archive.

Parameters

path (str) – Path where to save the container

Returns

savepath (str) – Path where the container was saved

to_hdf(*args, **kwargs)

Alias of Container().

classmethod load(pkid_or_path=None)

Load a container object from a persistent location or file path.

Parameters

pkid_or_path – Integer pkid corresponding to the container table or file path

Returns

container – The saved container object

classmethod from_hdf(*args, **kwargs)

Alias for Container().

class exa.core.container.TypedMeta(name, bases, clsdict)

Bases: type

This metaclass creates statically typed class attributes using the property framework.

class TestMeta(TypedMeta):
    attr1 = (int, float)
    attr2 = DataFrame

class TestClass(metaclass=TestMeta):
    def __init__(self, attr1, attr2):
        self.attr1 = attr1
        self.attr2 = attr2

The above code dynamically creates code that looks like the following:

class TestClass:
    @property
    def attr1(self):
        return self._attr1

    @attr1.setter
    def attr1(self, obj):
        if not isinstance(obj, (int, float)):
            raise TypeError('attr1 must be int')
        self._attr1 = obj

    @attr1.deleter
    def attr1(self):
        del self._attr1

    @property
    def attr2(self):
        return self._attr2

    @attr2.setter
    def attr2(self, obj):
        if not isinstance(obj, DataFrame):
            raise TypeError('attr2 must be DataFrame')
        self._attr2 = obj

    @attr2.deleter
    def attr2(self):
        del self._attr2

    def __init__(self, attr1, attr2):
        self.attr1 = attr1
        self.attr2 = attr2

static create_property(name, ptype)

Creates a custom property with a getter that performs computing functionality (if available) and raise a type error if setting with the wrong type.

Note

By default, the setter attempts to convert the object to the correct type; a type error is raised if this fails.