You can run this notebook in a live session Binder or view it on Github.

DataFrames: Groupby

This notebook uses the Pandas groupby-aggregate and groupby-apply on scalable Dask dataframes. It will discuss both common use and best practices.

Start Dask Client for Dashboard

Starting the Dask Client is optional. It will provide a dashboard which is useful to gain insight on the computation.

The link to the dashboard will become visible when you create the client below. We recommend having it open on one side of your screen while using your notebook on the other side. This can take some effort to arrange your windows, but seeing them both at the same is very useful when learning.

In [1]:
from dask.distributed import Client
client = Client(n_workers=1, threads_per_worker=4, processes=False, memory_limit='2GB')
client
Out[1]:

Client

Cluster

  • Workers: 1
  • Cores: 4
  • Memory: 2.00 GB

Artifical dataset

We create an artificial timeseries dataset to help us work with groupby operations

In [2]:
import dask
df = dask.datasets.timeseries()
df
Out[2]:
Dask DataFrame Structure:
id name x y
npartitions=30
2000-01-01 int64 object float64 float64
2000-01-02 ... ... ... ...
... ... ... ... ...
2000-01-30 ... ... ... ...
2000-01-31 ... ... ... ...
Dask Name: make-timeseries, 30 tasks

This dataset is small enough to fit in the cluster’s memory, so we persist it now.

You would skip this step if your dataset becomes too large to fit into memory.

In [3]:
df = df.persist()

Groupby Aggregations

Dask dataframes implement a commonly used subset of the Pandas groupby API (see Pandas Groupby Documentation.

We start with groupby aggregations. These are generally fairly efficient, assuming that the number of groups is small (less than a million).

In [4]:
df.groupby('name').x.mean().compute()
Out[4]:
name
Alice      -0.003574
Bob        -0.004333
Charlie    -0.002944
Dan        -0.000717
Edith      -0.000791
Frank       0.000112
George     -0.002299
Hannah      0.001146
Ingrid      0.000887
Jerry       0.000448
Kevin      -0.000860
Laura       0.001426
Michael    -0.000779
Norbert    -0.003489
Oliver     -0.000035
Patricia   -0.002386
Quinn       0.001182
Ray         0.003122
Sarah       0.000034
Tim        -0.000152
Ursula     -0.001579
Victor     -0.001319
Wendy      -0.001101
Xavier     -0.000267
Yvonne      0.000073
Zelda       0.000965
Name: x, dtype: float64

Performance will depend on the aggregation you do (mean vs std), the key on which you group (name vs id), and the number of total groups

In [5]:
%time _ = df.groupby('id').x.mean().compute()
CPU times: user 380 ms, sys: 12 ms, total: 392 ms
Wall time: 293 ms
In [6]:
%time _ = df.groupby('name').x.mean().compute()
CPU times: user 708 ms, sys: 32 ms, total: 740 ms
Wall time: 524 ms
In [7]:
%time df.groupby('name').agg({'x': ['mean', 'std'], 'y': ['mean', 'count']}).compute().head()
CPU times: user 1.2 s, sys: 64 ms, total: 1.27 s
Wall time: 966 ms
Out[7]:
x y
mean std mean count
name
Alice -0.003574 0.577825 -0.000265 100187
Bob -0.004333 0.577542 0.000859 99430
Charlie -0.002944 0.577864 -0.000018 99684
Dan -0.000717 0.576092 -0.001146 99802
Edith -0.000791 0.577220 -0.001763 99987

This is the same as with Pandas. Generally speaking, Dask.dataframe groupby-aggregations are roughly same performance as Pandas groupby-aggregations, just more scalable.

You can read more about Pandas’ common aggregations in the Pandas documentation.

Many groups

By default groupby-aggregations (like groupby-mean or groupby-sum) return the result as a single-partition Dask dataframe. Their results are usually quite small, so this is usually a good choice.

However, sometimes people want to do groupby aggregations on many groups (millions or more). In these cases the full result may not fit into a single Pandas dataframe output, and you may need to split your output into multiple partitions. You can control this with the split_out= parameter

In [8]:
# Computational graph of a single output aggregation (for a small number of groups, like 1000)
df.groupby('name').x.mean().visualize(node_attr={'penwidth': '6'})
Out[8]:
../_images/dataframes_02-groupby_15_0.png
In [9]:
# Computational graph of an aggregation to four outputs (for a larger number of groups, like 1000000)
df.groupby('id').x.mean(split_out=4).visualize(node_attr={'penwidth': '6'})
dot: graph is too large for cairo-renderer bitmaps. Scaling by 0.717677 to fit
Out[9]:
../_images/dataframes_02-groupby_16_1.png

Groupby Apply

Groupby-aggregations are generally quite fast because they can be broken down easily into well known operations. The data doesn’t have to move around too much and we can just pass around small intermediate values across the network.

For some operations however the function to be applied requires all data from a given group (like every record of someone named “Alice”). This will force a great deal of communication and be more expensive, but is still possible with the Groupby-apply method. This should be avoided if a groupby-aggregation works.

In the following example we train a simple Scikit-Learn machine learning model on every person’s name.

In [10]:
from sklearn.linear_model import LinearRegression

def train(partition):
    est = LinearRegression()
    est.fit(partition[['x', 'id']].values, partition.y.values)
    return est
In [11]:
%time df.groupby('name').apply(train, meta=object).compute().sort_index()
CPU times: user 5.88 s, sys: 680 ms, total: 6.56 s
Wall time: 5.27 s
Out[11]:
name
Alice       LinearRegression(copy_X=True, fit_intercept=Tr...
Bob         LinearRegression(copy_X=True, fit_intercept=Tr...
Charlie     LinearRegression(copy_X=True, fit_intercept=Tr...
Dan         LinearRegression(copy_X=True, fit_intercept=Tr...
Edith       LinearRegression(copy_X=True, fit_intercept=Tr...
Frank       LinearRegression(copy_X=True, fit_intercept=Tr...
George      LinearRegression(copy_X=True, fit_intercept=Tr...
Hannah      LinearRegression(copy_X=True, fit_intercept=Tr...
Ingrid      LinearRegression(copy_X=True, fit_intercept=Tr...
Jerry       LinearRegression(copy_X=True, fit_intercept=Tr...
Kevin       LinearRegression(copy_X=True, fit_intercept=Tr...
Laura       LinearRegression(copy_X=True, fit_intercept=Tr...
Michael     LinearRegression(copy_X=True, fit_intercept=Tr...
Norbert     LinearRegression(copy_X=True, fit_intercept=Tr...
Oliver      LinearRegression(copy_X=True, fit_intercept=Tr...
Patricia    LinearRegression(copy_X=True, fit_intercept=Tr...
Quinn       LinearRegression(copy_X=True, fit_intercept=Tr...
Ray         LinearRegression(copy_X=True, fit_intercept=Tr...
Sarah       LinearRegression(copy_X=True, fit_intercept=Tr...
Tim         LinearRegression(copy_X=True, fit_intercept=Tr...
Ursula      LinearRegression(copy_X=True, fit_intercept=Tr...
Victor      LinearRegression(copy_X=True, fit_intercept=Tr...
Wendy       LinearRegression(copy_X=True, fit_intercept=Tr...
Xavier      LinearRegression(copy_X=True, fit_intercept=Tr...
Yvonne      LinearRegression(copy_X=True, fit_intercept=Tr...
Zelda       LinearRegression(copy_X=True, fit_intercept=Tr...
dtype: object