.. currentmodule:: pandas
.. ipython:: python :suppress: import numpy as np np.random.seed(123456) from numpy import nan from pandas import * randn = np.random.randn np.set_printoptions(precision=4, suppress=True)
pandas provides various facilities for easily combining together Series, DataFrame, and Panel objects with various kinds of set logic for the indexes and relational algebra functionality in the case of join / merge-type operations.
The concat function (in the main pandas namespace) does all of the heavy
lifting of performing concatenation operations along an axis while performing
optional set logic (union or intersection) of the indexes (if any) on the other
axes. Note that I say "if any" because there is only a single possible axis of
concatenation for Series.
Before diving into all of the details of concat and what it can do, here is
a simple example:
.. ipython:: python df = DataFrame(np.random.randn(10, 4)) df # break it into pieces pieces = [df[:3], df[3:7], df[7:]] concatenated = concat(pieces) concatenated
Like its sibling function on ndarrays, numpy.concatenate, pandas.concat
takes a list or dict of homogeneously-typed objects and concatenates them with
some configurable handling of "what to do with the other axes":
concat(objs, axis=0, join='outer', join_axes=None, ignore_index=False,
keys=None, levels=None, names=None, verify_integrity=False)
objs: list or dict of Series, DataFrame, or Panel objects. If a dict is passed, the sorted keys will be used as the keys argument, unless it is passed, in which case the values will be selected (see below)axis: {0, 1, ...}, default 0. The axis to concatenate alongjoin: {'inner', 'outer'}, default 'outer'. How to handle indexes on other axis(es). Outer for union and inner for intersectionjoin_axes: list of Index objects. Specific indexes to use for the other n - 1 axes instead of performing inner/outer set logickeys: sequence, default None. Construct hierarchical index using the passed keys as the outermost level If multiple levels passed, should contain tuples.levels: list of sequences, default None. If keys passed, specific levels to use for the resulting MultiIndex. Otherwise they will be inferred from the keysnames: list, default None. Names for the levels in the resulting hierarchical indexverify_integrity: boolean, default False. Check whether the new concatenated axis contains duplicates. This can be very expensive relative to the actual data concatenationignore_index: boolean, default False. If True, do not use the index values on the concatenation axis. The resulting axis will be labeled 0, ..., n - 1. This is useful if you are concatenating objects where the concatenation axis does not have meaningful indexing information.
Without a little bit of context and example many of these arguments don't make
much sense. Let's take the above example. Suppose we wanted to associate
specific keys with each of the pieces of the chopped up DataFrame. We can do
this using the keys argument:
.. ipython:: python concatenated = concat(pieces, keys=['first', 'second', 'third']) concatenated
As you can see (if you've read the rest of the documentation), the resulting object's index has a :ref:`hierarchical index <indexing.hierarchical>`. This means that we can now do stuff like select out each chunk by key:
.. ipython:: python concatenated.ix['second']
It's not a stretch to see how this can be very useful. More detail on this functionality below.
When gluing together multiple DataFrames (or Panels or...), for example, you have a choice of how to handle the other axes (other than the one being concatenated). This can be done in three ways:
- Take the (sorted) union of them all,
join='outer'. This is the default option as it results in zero information loss. - Take the intersection,
join='inner'. - Use a specific index (in the case of DataFrame) or indexes (in the case of
Panel or future higher dimensional objects), i.e. the
join_axesargument
Here is a example of each of these methods. First, the default join='outer'
behavior:
.. ipython:: python
from pandas.util.testing import rands
df = DataFrame(np.random.randn(10, 4), columns=['a', 'b', 'c', 'd'],
index=[rands(5) for _ in range(10)])
df
concat([df.ix[:7, ['a', 'b']], df.ix[2:-2, ['c']],
df.ix[-7:, ['d']]], axis=1)
Note that the row indexes have been unioned and sorted. Here is the same thing
with join='inner':
.. ipython:: python
concat([df.ix[:7, ['a', 'b']], df.ix[2:-2, ['c']],
df.ix[-7:, ['d']]], axis=1, join='inner')
Lastly, suppose we just wanted to reuse the exact index from the original DataFrame:
.. ipython:: python
concat([df.ix[:7, ['a', 'b']], df.ix[2:-2, ['c']],
df.ix[-7:, ['d']]], axis=1, join_axes=[df.index])
A useful shortcut to concat are the append instance methods on Series
and DataFrame. These methods actually predated concat. They concatenate
along axis=0, namely the index:
.. ipython:: python s = Series(randn(10), index=np.arange(10)) s1 = s[:5] # note we're slicing with labels here, so 5 is included s2 = s[6:] s1.append(s2)
In the case of DataFrame, the indexes must be disjoint but the columns do not need to be:
.. ipython:: python
df = DataFrame(randn(6, 4), index=date_range('1/1/2000', periods=6),
columns=['A', 'B', 'C', 'D'])
df1 = df.ix[:3]
df2 = df.ix[3:, :3]
df1
df2
df1.append(df2)
append may take multiple objects to concatenate:
.. ipython:: python df1 = df.ix[:2] df2 = df.ix[2:4] df3 = df.ix[4:] df1.append([df2,df3])
Note
Unlike list.append method, which appends to the original list and
returns nothing, append here does not modify df1 and
returns its copy with df2 appended.
For DataFrames which don't have a meaningful index, you may wish to append them and ignore the fact that they may have overlapping indexes:
.. ipython:: python df1 = DataFrame(randn(6, 4), columns=['A', 'B', 'C', 'D']) df2 = DataFrame(randn(3, 4), columns=['A', 'B', 'C', 'D']) df1 df2
To do this, use the ignore_index argument:
.. ipython:: python concat([df1, df2], ignore_index=True)
This is also a valid argument to DataFrame.append:
.. ipython:: python df1.append(df2, ignore_index=True)
Let's consider a variation on the first example presented:
.. ipython:: python df = DataFrame(np.random.randn(10, 4)) df # break it into pieces pieces = [df.ix[:, [0, 1]], df.ix[:, [2]], df.ix[:, [3]]] result = concat(pieces, axis=1, keys=['one', 'two', 'three']) result
You can also pass a dict to concat in which case the dict keys will be used
for the keys argument (unless other keys are specified):
.. ipython:: python
pieces = {'one': df.ix[:, [0, 1]],
'two': df.ix[:, [2]],
'three': df.ix[:, [3]]}
concat(pieces, axis=1)
concat(pieces, keys=['three', 'two'])
The MultiIndex created has levels that are constructed from the passed keys and the columns of the DataFrame pieces:
.. ipython:: python result.columns.levels
If you wish to specify other levels (as will occasionally be the case), you can
do so using the levels argument:
.. ipython:: python
result = concat(pieces, axis=1, keys=['one', 'two', 'three'],
levels=[['three', 'two', 'one', 'zero']],
names=['group_key'])
result
result.columns.levels
Yes, this is fairly esoteric, but is actually necessary for implementing things like GroupBy where the order of a categorical variable is meaningful.
While not especially efficient (since a new object must be created), you can
append a single row to a DataFrame by passing a Series or dict to append,
which returns a new DataFrame as above.
.. ipython:: python df = DataFrame(np.random.randn(8, 4), columns=['A','B','C','D']) df s = df.xs(3) df.append(s, ignore_index=True)
You should use ignore_index with this method to instruct DataFrame to
discard its index. If you wish to preserve the index, you should construct an
appropriately-indexed DataFrame and append or concatenate those objects.
You can also pass a list of dicts or Series:
.. ipython:: python
df = DataFrame(np.random.randn(5, 4),
columns=['foo', 'bar', 'baz', 'qux'])
dicts = [{'foo': 1, 'bar': 2, 'baz': 3, 'peekaboo': 4},
{'foo': 5, 'bar': 6, 'baz': 7, 'peekaboo': 8}]
result = df.append(dicts, ignore_index=True)
result
pandas has full-featured, high performance in-memory join operations
idiomatically very similar to relational databases like SQL. These methods
perform significantly better (in some cases well over an order of magnitude
better) than other open source implementations (like base::merge.data.frame
in R). The reason for this is careful algorithmic design and internal layout of
the data in DataFrame.
See the :ref:`cookbook<cookbook.merge>` for some advanced strategies
pandas provides a single function, merge, as the entry point for all
standard database join operations between DataFrame objects:
merge(left, right, how='left', on=None, left_on=None, right_on=None,
left_index=False, right_index=False, sort=True,
suffixes=('_x', '_y'), copy=True)
Here's a description of what each argument is for:
left: A DataFrame objectright: Another DataFrame objecton: Columns (names) to join on. Must be found in both the left and right DataFrame objects. If not passed andleft_indexandright_indexareFalse, the intersection of the columns in the DataFrames will be inferred to be the join keysleft_on: Columns from the left DataFrame to use as keys. Can either be column names or arrays with length equal to the length of the DataFrameright_on: Columns from the right DataFrame to use as keys. Can either be column names or arrays with length equal to the length of the DataFrameleft_index: IfTrue, use the index (row labels) from the left DataFrame as its join key(s). In the case of a DataFrame with a MultiIndex (hierarchical), the number of levels must match the number of join keys from the right DataFrameright_index: Same usage asleft_indexfor the right DataFramehow: One of'left','right','outer','inner'. Defaults toinner. See below for more detailed description of each methodsort: Sort the result DataFrame by the join keys in lexicographical order. Defaults toTrue, setting toFalsewill improve performance substantially in many casessuffixes: A tuple of string suffixes to apply to overlapping columns. Defaults to('_x', '_y').copy: Always copy data (defaultTrue) from the passed DataFrame objects, even when reindexing is not necessary. Cannot be avoided in many cases but may improve performance / memory usage. The cases where copying can be avoided are somewhat pathological but this option is provided nonetheless.
merge is a function in the pandas namespace, and it is also available as a
DataFrame instance method, with the calling DataFrame being implicitly
considered the left object in the join.
The related DataFrame.join method, uses merge internally for the
index-on-index and index-on-column(s) joins, but joins on indexes by default
rather than trying to join on common columns (the default behavior for
merge). If you are joining on index, you may wish to use DataFrame.join
to save yourself some typing.
Experienced users of relational databases like SQL will be familiar with the terminology used to describe join operations between two SQL-table like structures (DataFrame objects). There are several cases to consider which are very important to understand:
- one-to-one joins: for example when joining two DataFrame objects on their indexes (which must contain unique values)
- many-to-one joins: for example when joining an index (unique) to one or more columns in a DataFrame
- many-to-many joins: joining columns on columns.
Note
When joining columns on columns (potentially a many-to-many join), any indexes on the passed DataFrame objects will be discarded.
It is worth spending some time understanding the result of the many-to-many join case. In SQL / standard relational algebra, if a key combination appears more than once in both tables, the resulting table will have the Cartesian product of the associated data. Here is a very basic example with one unique key combination:
.. ipython:: python
left = DataFrame({'key': ['foo', 'foo'], 'lval': [1, 2]})
right = DataFrame({'key': ['foo', 'foo'], 'rval': [4, 5]})
left
right
merge(left, right, on='key')
Here is a more complicated example with multiple join keys:
.. ipython:: python
left = DataFrame({'key1': ['foo', 'foo', 'bar'],
'key2': ['one', 'two', 'one'],
'lval': [1, 2, 3]})
right = DataFrame({'key1': ['foo', 'foo', 'bar', 'bar'],
'key2': ['one', 'one', 'one', 'two'],
'rval': [4, 5, 6, 7]})
merge(left, right, how='outer')
merge(left, right, how='inner')
The how argument to merge specifies how to determine which keys are to
be included in the resulting table. If a key combination does not appear in
either the left or right tables, the values in the joined table will be
NA. Here is a summary of the how options and their SQL equivalent names:
| Merge method | SQL Join Name | Description |
|---|---|---|
left |
LEFT OUTER JOIN |
Use keys from left frame only |
right |
RIGHT OUTER JOIN |
Use keys from right frame only |
outer |
FULL OUTER JOIN |
Use union of keys from both frames |
inner |
INNER JOIN |
Use intersection of keys from both frames |
DataFrame.join is a convenient method for combining the columns of two
potentially differently-indexed DataFrames into a single result DataFrame. Here
is a very basic example:
.. ipython:: python df = DataFrame(np.random.randn(8, 4), columns=['A','B','C','D']) df1 = df.ix[1:, ['A', 'B']] df2 = df.ix[:5, ['C', 'D']] df1 df2 df1.join(df2) df1.join(df2, how='outer') df1.join(df2, how='inner')
The data alignment here is on the indexes (row labels). This same behavior can
be achieved using merge plus additional arguments instructing it to use the
indexes:
.. ipython:: python merge(df1, df2, left_index=True, right_index=True, how='outer')
join takes an optional on argument which may be a column or multiple
column names, which specifies that the passed DataFrame is to be aligned on
that column in the DataFrame. These two function calls are completely
equivalent:
left.join(right, on=key_or_keys)
merge(left, right, left_on=key_or_keys, right_index=True,
how='left', sort=False)
Obviously you can choose whichever form you find more convenient. For
many-to-one joins (where one of the DataFrame's is already indexed by the join
key), using join may be more convenient. Here is a simple example:
.. ipython:: python
df['key'] = ['foo', 'bar'] * 4
to_join = DataFrame(randn(2, 2), index=['bar', 'foo'],
columns=['j1', 'j2'])
df
to_join
df.join(to_join, on='key')
merge(df, to_join, left_on='key', right_index=True,
how='left', sort=False)
To join on multiple keys, the passed DataFrame must have a MultiIndex:
.. ipython:: python
index = MultiIndex(levels=[['foo', 'bar', 'baz', 'qux'],
['one', 'two', 'three']],
labels=[[0, 0, 0, 1, 1, 2, 2, 3, 3, 3],
[0, 1, 2, 0, 1, 1, 2, 0, 1, 2]],
names=['first', 'second'])
to_join = DataFrame(np.random.randn(10, 3), index=index,
columns=['j_one', 'j_two', 'j_three'])
# a little relevant example with NAs
key1 = ['bar', 'bar', 'bar', 'foo', 'foo', 'baz', 'baz', 'qux',
'qux', 'snap']
key2 = ['two', 'one', 'three', 'one', 'two', 'one', 'two', 'two',
'three', 'one']
data = np.random.randn(len(key1))
data = DataFrame({'key1' : key1, 'key2' : key2,
'data' : data})
data
to_join
Now this can be joined by passing the two key column names:
.. ipython:: python data.join(to_join, on=['key1', 'key2'])
The default for DataFrame.join is to perform a left join (essentially a
"VLOOKUP" operation, for Excel users), which uses only the keys found in the
calling DataFrame. Other join types, for example inner join, can be just as
easily performed:
.. ipython:: python data.join(to_join, on=['key1', 'key2'], how='inner')
As you can see, this drops any rows where there was no match.
The merge suffixes argument takes a tuple of list of strings to append to
overlapping column names in the input DataFrames to disambiguate the result
columns:
.. ipython:: python
left = DataFrame({'key': ['foo', 'foo'], 'value': [1, 2]})
right = DataFrame({'key': ['foo', 'foo'], 'value': [4, 5]})
merge(left, right, on='key', suffixes=['_left', '_right'])
DataFrame.join has lsuffix and rsuffix arguments which behave
similarly.
New in v0.8.0 is the ordered_merge function for combining time series and other
ordered data. In particular it has an optional fill_method keyword to
fill/interpolate missing data:
.. ipython:: python
:suppress:
A = DataFrame({'key' : ['a', 'c', 'e'] * 2,
'lvalue' : [1, 2, 3] * 2,
'group' : ['a', 'a', 'a', 'b', 'b', 'b']})
B = DataFrame({'key' : ['b', 'c', 'd'],
'rvalue' : [1, 2, 3]})
.. ipython:: python A B ordered_merge(A, B, fill_method='ffill', left_by='group')
A list or tuple of DataFrames can also be passed to DataFrame.join to join
them together on their indexes. The same is true for Panel.join.
.. ipython:: python df1 = df.ix[:, ['A', 'B']] df2 = df.ix[:, ['C', 'D']] df3 = df.ix[:, ['key']] df1 df1.join([df2, df3])
Another fairly common situation is to have two like-indexed (or similarly indexed) Series or DataFrame objects and wanting to "patch" values in one object from values for matching indices in the other. Here is an example:
.. ipython:: python
df1 = DataFrame([[nan, 3., 5.], [-4.6, np.nan, nan],
[nan, 7., nan]])
df2 = DataFrame([[-42.6, np.nan, -8.2], [-5., 1.6, 4]],
index=[1, 2])
For this, use the combine_first method:
.. ipython:: python df1.combine_first(df2)
Note that this method only takes values from the right DataFrame if they are
missing in the left DataFrame. A related method, update, alters non-NA
values inplace:
.. ipython:: python df1.update(df2) df1