pyspark.Accumulator¶

class pyspark.Accumulator(aid: int, value: T, accum_param: pyspark.accumulators.AccumulatorParam[T])¶

A shared variable that can be accumulated, i.e., has a commutative and associative “add” operation. Worker tasks on a Spark cluster can add values to an Accumulator with the += operator, but only the driver program is allowed to access its value, using value. Updates from the workers get propagated automatically to the driver program.

While SparkContext supports accumulators for primitive data types like int and float, users can also define accumulators for custom types by providing a custom AccumulatorParam object. Refer to its doctest for an example.

Examples

>>> a = sc.accumulator(1)
>>> a.value
1
>>> a.value = 2
>>> a.value
2
>>> a += 5
>>> a.value
7
>>> sc.accumulator(1.0).value
1.0
>>> sc.accumulator(1j).value
1j
>>> rdd = sc.parallelize([1,2,3])
>>> def f(x):
...     global a
...     a += x
>>> rdd.foreach(f)
>>> a.value
13
>>> b = sc.accumulator(0)
>>> def g(x):
...     b.add(x)
>>> rdd.foreach(g)
>>> b.value
6

>>> rdd.map(lambda x: a.value).collect() 
Traceback (most recent call last):
    ...
Py4JJavaError: ...

>>> def h(x):
...     global a
...     a.value = 7
>>> rdd.foreach(h) 
Traceback (most recent call last):
    ...
Py4JJavaError: ...

>>> sc.accumulator([1.0, 2.0, 3.0]) 
Traceback (most recent call last):
    ...
TypeError: ...

Methods

add(term)

Adds a term to this accumulator’s value

Attributes

value

Get the accumulator’s value; only usable in driver program

pyspark.Broadcast

pyspark.AccumulatorParam