API Reference

API Reference¶

Dask APIs generally follow from upstream APIs:

Arrays follows NumPy
DataFrames follows Pandas
Bag follows map/filter/groupby/reduce common in Spark and Python iterators
Delayed wraps general Python code
Futures follows concurrent.futures from the standard library for real-time computation.

Additionally, Dask has its own functions to start computations, persist data in memory, check progress, and so forth that complement the APIs above. These more general Dask functions are described below:

`compute`(*args[, traverse, optimize_graph, ...])	Compute several dask collections at once.
`is_dask_collection`(x)	Returns `True` if `x` is a dask collection.
`optimize`(*args[, traverse])	Optimize several dask collections at once.
`persist`(*args[, traverse, optimize_graph, ...])	Persist multiple Dask collections into memory
`visualize`(*args[, filename, traverse, ...])	Visualize several dask graphs simultaneously.

These functions work with any scheduler. More advanced operations are available when using the newer scheduler and starting a dask.distributed.Client (which, despite its name, runs nicely on a single machine). This API provides the ability to submit, cancel, and track work asynchronously, and includes many functions for complex inter-task workflows. These are not necessary for normal operation, but can be useful for real-time or advanced operation.

This more advanced API is available in the Dask distributed documentation

dask.annotate(**annotations: Any) → collections.abc.Iterator[None][source]¶

Context Manager for setting HighLevelGraph Layer annotations.

Annotations are metadata or soft constraints associated with tasks that dask schedulers may choose to respect: They signal intent without enforcing hard constraints. As such, they are primarily designed for use with the distributed scheduler.

Almost any object can serve as an annotation, but small Python objects are preferred, while large objects such as NumPy arrays are discouraged.

Callables supplied as an annotation should take a single key argument and produce the appropriate annotation. Individual task keys in the annotated collection are supplied to the callable.

Parameters

**annotationskey-value pairs

See also

annotate

dask.compute(*args, traverse=True, optimize_graph=True, scheduler=None, get=None, **kwargs)[source]¶

Compute several dask collections at once.

Parameters

argsobject: Any number of objects. If it is a dask object, it’s computed and the result is returned. By default, python builtin collections are also traversed to look for dask objects (for more information see the traverse keyword). Non-dask arguments are passed through unchanged.
traversebool, optional: By default dask traverses builtin python collections looking for dask objects passed to compute. For large collections this can be expensive. If none of the arguments contain any dask objects, set traverse=False to avoid doing this traversal.
schedulerstring, optional: Which scheduler to use like “threads”, “synchronous” or “processes”. If not provided, the default is to check the global settings first, and then fall back to the collection defaults.
optimize_graphbool, optional: If True [default], the optimizations for each collection are applied before computation. Otherwise the graph is run as is. This can be useful for debugging.
getNone: Should be left to None The get= keyword has been removed.
kwargs: Extra keywords to forward to the scheduler function.

Examples

>>> import dask
>>> import dask.array as da
>>> a = da.arange(10, chunks=2).sum()
>>> b = da.arange(10, chunks=2).mean()
>>> dask.compute(a, b)
(45, 4.5)

By default, dask objects inside python collections will also be computed:

>>> dask.compute({'a': a, 'b': b, 'c': 1})
({'a': 45, 'b': 4.5, 'c': 1},)

dask.is_dask_collection(x) → bool[source]¶

Returns True if x is a dask collection.

Parameters

xAny: Object to test.

Returns

resultbool: True if x is a Dask collection.

Notes

The DaskCollection typing.Protocol implementation defines a Dask collection as a class that returns a Mapping from the __dask_graph__ method. This helper function existed before the implementation of the protocol.

dask.optimize(*args, traverse=True, **kwargs)[source]¶

Optimize several dask collections at once.

Returns equivalent dask collections that all share the same merged and optimized underlying graph. This can be useful if converting multiple collections to delayed objects, or to manually apply the optimizations at strategic points.

Note that in most cases you shouldn’t need to call this method directly.

Parameters

*argsobjects: Any number of objects. If a dask object, its graph is optimized and merged with all those of all other dask objects before returning an equivalent dask collection. Non-dask arguments are passed through unchanged.
traversebool, optional: By default dask traverses builtin python collections looking for dask objects passed to optimize. For large collections this can be expensive. If none of the arguments contain any dask objects, set traverse=False to avoid doing this traversal.
optimizationslist of callables, optional: Additional optimization passes to perform.
**kwargs: Extra keyword arguments to forward to the optimization passes.

Examples

>>> import dask
>>> import dask.array as da
>>> a = da.arange(10, chunks=2).sum()
>>> b = da.arange(10, chunks=2).mean()
>>> a2, b2 = dask.optimize(a, b)

>>> a2.compute() == a.compute()
True
>>> b2.compute() == b.compute()
True

dask.persist(*args, traverse=True, optimize_graph=True, scheduler=None, **kwargs)[source]¶

Persist multiple Dask collections into memory

This turns lazy Dask collections into Dask collections with the same metadata, but now with their results fully computed or actively computing in the background.

For example a lazy dask.array built up from many lazy calls will now be a dask.array of the same shape, dtype, chunks, etc., but now with all of those previously lazy tasks either computed in memory as many small numpy.array (in the single-machine case) or asynchronously running in the background on a cluster (in the distributed case).

This function operates differently if a dask.distributed.Client exists and is connected to a distributed scheduler. In this case this function will return as soon as the task graph has been submitted to the cluster, but before the computations have completed. Computations will continue asynchronously in the background. When using this function with the single machine scheduler it blocks until the computations have finished.

When using Dask on a single machine you should ensure that the dataset fits entirely within memory.

Parameters

*args: Dask collections
schedulerstring, optional: Which scheduler to use like “threads”, “synchronous” or “processes”. If not provided, the default is to check the global settings first, and then fall back to the collection defaults.
traversebool, optional: By default dask traverses builtin python collections looking for dask objects passed to persist. For large collections this can be expensive. If none of the arguments contain any dask objects, set traverse=False to avoid doing this traversal.
optimize_graphbool, optional: If True [default], the graph is optimized before computation. Otherwise the graph is run as is. This can be useful for debugging.
**kwargs: Extra keywords to forward to the scheduler function.

Returns

New dask collections backed by in-memory data

Examples

>>> df = dd.read_csv('/path/to/*.csv')  
>>> df = df[df.name == 'Alice']  
>>> df['in-debt'] = df.balance < 0  
>>> df = df.persist()  # triggers computation  

>>> df.value().min()  # future computations are now fast  
-10
>>> df.value().max()  
100

>>> from dask import persist  # use persist function on multiple collections
>>> a, b = persist(a, b)  

dask.visualize(*args, filename='mydask', traverse=True, optimize_graph=False, maxval=None, engine: Optional[Literal['cytoscape', 'ipycytoscape', 'graphviz']] = None, **kwargs)[source]¶

Visualize several dask graphs simultaneously.

Requires graphviz to be installed. All options that are not the dask graph(s) should be passed as keyword arguments.

Parameters

argsobject

Any number of objects. If it is a dask collection (for example, a dask DataFrame, Array, Bag, or Delayed), its associated graph will be included in the output of visualize. By default, python builtin collections are also traversed to look for dask objects (for more information see the traverse keyword). Arguments lacking an associated graph will be ignored.

filenamestr or None, optional

The name of the file to write to disk. If the provided filename doesn’t include an extension, ‘.png’ will be used by default. If filename is None, no file will be written, and we communicate with dot using only pipes.

format{‘png’, ‘pdf’, ‘dot’, ‘svg’, ‘jpeg’, ‘jpg’}, optional

Format in which to write output file. Default is ‘png’.

traversebool, optional

By default, dask traverses builtin python collections looking for dask objects passed to visualize. For large collections this can be expensive. If none of the arguments contain any dask objects, set traverse=False to avoid doing this traversal.

optimize_graphbool, optional

If True, the graph is optimized before rendering. Otherwise, the graph is displayed as is. Default is False.

color{None, ‘order’, ‘ages’, ‘freed’, ‘memoryincreases’, ‘memorydecreases’, ‘memorypressure’}, optional

Options to color nodes. colormap:

None, the default, no colors.
‘order’, colors the nodes’ border based on the order they appear in the graph.
‘ages’, how long the data of a node is held.
‘freed’, the number of dependencies released after running a node.
‘memoryincreases’, how many more outputs are held after the lifetime of a node. Large values may indicate nodes that should have run later.
‘memorydecreases’, how many fewer outputs are held after the lifetime of a node. Large values may indicate nodes that should have run sooner.
‘memorypressure’, the number of data held when the node is run (circle), or the data is released (rectangle).

maxval{int, float}, optional

Maximum value for colormap to normalize form 0 to 1.0. Default is None will make it the max number of values

collapse_outputsbool, optional

Whether to collapse output boxes, which often have empty labels. Default is False.

verbosebool, optional

Whether to label output and input boxes even if the data aren’t chunked. Beware: these labels can get very long. Default is False.

engine{“graphviz”, “ipycytoscape”, “cytoscape”}, optional.

The visualization engine to use. If not provided, this checks the dask config value “visualization.engine”. If that is not set, it tries to import graphviz and ipycytoscape, using the first one to succeed.

**kwargs

Additional keyword arguments to forward to the visualization engine.

Returns

resultIPython.diplay.Image, IPython.display.SVG, or None: See dask.dot.dot_graph for more information.

Datasets¶

Dask has a few helpers for generating demo datasets

dask.datasets.make_people(npartitions=10, records_per_partition=1000, seed=None, locale='en')[source]¶

Make a dataset of random people

This makes a Dask Bag with dictionary records of randomly generated people. This requires the optional library mimesis to generate records.

Parameters

npartitionsint: Number of partitions
records_per_partitionint: Number of records in each partition
seedint, (optional): Random seed
localestr: Language locale, like ‘en’, ‘fr’, ‘zh’, or ‘ru’

Returns

b: Dask Bag

dask.datasets.timeseries(start='2000-01-01', end='2000-01-31', freq='1s', partition_freq='1d', dtypes=None, seed=None, **kwargs)[source]¶

Create timeseries dataframe with random data

Parameters

startdatetime (or datetime-like string): Start of time series
enddatetime (or datetime-like string): End of time series
dtypesdict (optional): Mapping of column names to types. Valid types include {float, int, str, ‘category’}
freqstring: String like ‘2s’ or ‘1H’ or ‘12W’ for the time series frequency
partition_freqstring: String like ‘1M’ or ‘2Y’ to divide the dataframe into partitions
seedint (optional): Randomstate seed
kwargs:: Keywords to pass down to individual column creation functions. Keywords should be prefixed by the column name and then an underscore.

Examples

>>> import dask
>>> df = dask.datasets.timeseries()
>>> df.head()  
          timestamp    id     name         x         y
2000-01-01 00:00:00   967    Jerry -0.031348 -0.040633
2000-01-01 00:00:01  1066  Michael -0.262136  0.307107
2000-01-01 00:00:02   988    Wendy -0.526331  0.128641
2000-01-01 00:00:03  1016   Yvonne  0.620456  0.767270
2000-01-01 00:00:04   998   Ursula  0.684902 -0.463278
>>> df = dask.datasets.timeseries(
...     '2000', '2010',
...     freq='2H', partition_freq='1D', seed=1,  # data frequency
...     dtypes={'value': float, 'name': str, 'id': int},  # data types
...     id_lam=1000  # control number of items in id column
... )

Datasets with defined specs¶

The following helpers are still experimental:

dask.dataframe.io.demo.with_spec(spec: dask.dataframe.io.demo.DatasetSpec, seed: int | None = None)[source]¶

Generate a random dataset according to provided spec

Parameters

specDatasetSpec: Specify all the parameters of the dataset
seed: int (optional): Randomstate seed

Notes

This API is still experimental, and will likely change in the future

Examples

>>> from dask.dataframe.io.demo import ColumnSpec, DatasetSpec, with_spec
>>> ddf = with_spec(
...     DatasetSpec(
...         npartitions=10,
...         nrecords=10_000,
...         column_specs=[
...             ColumnSpec(dtype=int, number=2, prefix="p"),
...             ColumnSpec(dtype=int, number=2, prefix="n", method="normal"),
...             ColumnSpec(dtype=float, number=2, prefix="f"),
...             ColumnSpec(dtype=str, prefix="s", number=2, random=True, length=10),
...             ColumnSpec(dtype="category", prefix="c", choices=["Y", "N"]),
...         ],
...     ), seed=42)
>>> ddf.head(10)  
     p1    p2    n1    n2        f1        f2          s1          s2 c1
0  1002   972  -811    20  0.640846 -0.176875  L#h98#}J`?  _8C607/:6e  N
1   985   982 -1663  -777  0.790257  0.792796  u:XI3,omoZ  w~@ /d)'-@  N
2   947   970   799  -269  0.740869 -0.118413  O$dnwCuq\  !WtSe+(;#9  Y
3  1003   983  1133   521 -0.987459  0.278154  j+Qr_2{XG&  &XV7cy$y1T  Y
4  1017  1049   826     5 -0.875667 -0.744359  bJ3E-{:o  {+jC).?vK+  Y
5   984  1017  -492  -399  0.748181  0.293761  ~zUNHNgD"!  yuEkXeVot|  Y
6   992  1027  -856    67 -0.125132 -0.234529  j.7z;o]Gc9  g|Fi5*}Y92  Y
7  1011   974   762 -1223  0.471696  0.937935  yT?j~N/-u]  JhEB[W-}^$  N
8   984   974   856    74  0.109963  0.367864  _j"&@ i&;/  OYXQ)w{hoH  N
9  1030  1001  -792  -262  0.435587 -0.647970  Pmrwl{{|.K  3UTqM$86Sg  N

The `ColumnSpec` class¶

class dask.dataframe.io.demo.ColumnSpec(prefix: str | None = None, dtype: str | type | None = None, number: int = 1, nunique: int | None = None, choices: list = <factory>, low: int | None = None, high: int | None = None, length: int | None = None, random: bool = False, method: str | None = None, args: tuple[typing.Any, ...] = <factory>, kwargs: dict[str, typing.Any] = <factory>)[source]¶

Bases: object

Encapsulates properties of a family of columns with the same dtype. Different method can be specified for integer dtype (“poisson”, “uniform”, “binomial”, etc.)

Notes

This API is still experimental, and will likely change in the future

args: tuple[Any, ...]¶: Args to pass into the method

choices: list¶: For a “category” or str column, list of possible values

dtype: str | type | None = None¶: Column data type. Only supports numpy dtypes

high: int | None = None¶: For an int column, high end of range

kwargs: dict[str, Any]¶: Any other kwargs to pass into the method

length: int | None = None¶: For a str or “category” column with random=True, how large a string to generate

low: int | None = None¶: Start value for an int column. Optional if random=True, since randint doesn’t accept high and low.

method: str | None = None¶: For an int column, method to use when generating the value, such as “poisson”, “uniform”, “binomial”. Default “poisson”. Delegates to the same method of RandomState

number: int = 1¶: How many columns to create with these properties. Default 1. If more than one columns are specified, they will be numbered: “int1”, “int2”, etc.

nunique: int | None = None¶: For a “category” column, how many unique categories to generate

prefix: str | None = None¶: Column prefix. If not specified, will default to str(dtype)

random: bool = False¶: For an int column, whether to use randint. For a string column produces a random string of specified length

The `RangeIndexSpec` class¶

class dask.dataframe.io.demo.RangeIndexSpec(dtype: str | type = <class 'int'>, step: int = 1)[source]¶

Bases: object

Properties of the dataframe RangeIndex

Notes

This API is still experimental, and will likely change in the future

dtype¶

Index dtype

alias of int

step: int = 1¶: Step for a RangeIndex

The `DatetimeIndexSpec` class¶

class dask.dataframe.io.demo.DatetimeIndexSpec(dtype: str | type = <class 'int'>, start: str | None = None, freq: str = '1H', partition_freq: str | None = None)[source]¶

Bases: object

Properties of the dataframe DatetimeIndex

Notes

This API is still experimental, and will likely change in the future

dtype¶

Index dtype

alias of int

freq: str = '1H'¶: Frequency for the index (“1H”, “1D”, etc.)

partition_freq: str | None = None¶: Partition frequency (“1D”, “1M”, etc.)

start: str | None = None¶: First value of the index

The `DatasetSpec` class¶

class dask.dataframe.io.demo.DatasetSpec(npartitions: int = 1, nrecords: int = 1000, index_spec: dask.dataframe.io.demo.RangeIndexSpec | dask.dataframe.io.demo.DatetimeIndexSpec = <factory>, column_specs: list[dask.dataframe.io.demo.ColumnSpec] = <factory>)[source]¶

Bases: object

Defines a dataset with random data, such as which columns and data types to generate

Notes

This API is still experimental, and will likely change in the future

column_specs: list[dask.dataframe.io.demo.ColumnSpec]¶: List of column definitions

index_spec: dask.dataframe.io.demo.RangeIndexSpec | dask.dataframe.io.demo.DatetimeIndexSpec¶: Properties of the index

npartitions: int = 1¶: How many partitions generate in the dataframe. If the dataframe has a DatetimeIndex, specify its partition_freq instead

nrecords: int = 1000¶: Total number of records to generate

Utilities¶

Dask has some public utility methods. These are primarily used for parsing configuration values.

dask.utils.apply(func, args, kwargs=None)[source]¶

Apply a function given its positional and keyword arguments.

Equivalent to func(*args, **kwargs) Most Dask users will never need to use the apply function. It is typically only used by people who need to inject keyword argument values into a low level Dask task graph.

Parameters

funccallable: The function you want to apply.
argstuple: A tuple containing all the positional arguments needed for func (eg: (arg_1, arg_2, arg_3))
kwargsdict, optional: A dictionary mapping the keyword arguments (eg: {"kwarg_1": value, "kwarg_2": value}

Examples

>>> from dask.utils import apply
>>> def add(number, second_number=5):
...     return number + second_number
...
>>> apply(add, (10,), {"second_number": 2})  # equivalent to add(*args, **kwargs)
12

>>> task = apply(add, (10,), {"second_number": 2})
>>> dsk = {'task-name': task}  # adds the task to a low level Dask task graph

dask.utils.format_bytes(n: int) → str[source]¶

Format bytes as text

>>> from dask.utils import format_bytes
>>> format_bytes(1)
'1 B'
>>> format_bytes(1234)
'1.21 kiB'
>>> format_bytes(12345678)
'11.77 MiB'
>>> format_bytes(1234567890)
'1.15 GiB'
>>> format_bytes(1234567890000)
'1.12 TiB'
>>> format_bytes(1234567890000000)
'1.10 PiB'

For all values < 2**60, the output is always <= 10 characters.

dask.utils.format_time(n: float) → str[source]¶

format integers as time

>>> from dask.utils import format_time
>>> format_time(1)
'1.00 s'
>>> format_time(0.001234)
'1.23 ms'
>>> format_time(0.00012345)
'123.45 us'
>>> format_time(123.456)
'123.46 s'
>>> format_time(1234.567)
'20m 34s'
>>> format_time(12345.67)
'3hr 25m'
>>> format_time(123456.78)
'34hr 17m'
>>> format_time(1234567.89)
'14d 6hr'

dask.utils.parse_bytes(s: float | str) → int[source]¶

Parse byte string to numbers

>>> from dask.utils import parse_bytes
>>> parse_bytes('100')
100
>>> parse_bytes('100 MB')
100000000
>>> parse_bytes('100M')
100000000
>>> parse_bytes('5kB')
5000
>>> parse_bytes('5.4 kB')
5400
>>> parse_bytes('1kiB')
1024
>>> parse_bytes('1e6')
1000000
>>> parse_bytes('1e6 kB')
1000000000
>>> parse_bytes('MB')
1000000
>>> parse_bytes(123)
123
>>> parse_bytes('5 foos')
Traceback (most recent call last):
    ...
ValueError: Could not interpret 'foos' as a byte unit

dask.utils.parse_timedelta(s: None, default: Union[str, Literal[False]] = 'seconds') → None[source]¶

dask.utils.parse_timedelta(s: str | float | datetime.timedelta, default: Union[str, Literal[False]] = 'seconds') → float

Parse timedelta string to number of seconds

Parameters

sstr, float, timedelta, or None
default: str or False, optional: Unit of measure if s does not specify one. Defaults to seconds. Set to False to require s to explicitly specify its own unit.

Examples

>>> from datetime import timedelta
>>> from dask.utils import parse_timedelta
>>> parse_timedelta('3s')
3
>>> parse_timedelta('3.5 seconds')
3.5
>>> parse_timedelta('300ms')
0.3
>>> parse_timedelta(timedelta(seconds=3))  # also supports timedeltas
3

Scheduling in Depth

API

API Reference

Contents

API Reference¶

Datasets¶

Datasets with defined specs¶

The ColumnSpec class¶

The RangeIndexSpec class¶

The DatetimeIndexSpec class¶

The DatasetSpec class¶

Utilities¶

The `ColumnSpec` class¶

The `RangeIndexSpec` class¶

The `DatetimeIndexSpec` class¶

The `DatasetSpec` class¶