Intro to Data Structures¶
We’ll start with a quick, non-comprehensive overview of the fundamental data structures in pandas to get you started. The fundamental behavior about data types, indexing, and axis labeling / alignment apply across all of the objects. To get started, import numpy and load pandas into your namespace:
In [451]: import numpy as np
# will use a lot in examples
In [452]: randn = np.random.randn
In [453]: from pandas import *
Here is a basic tenet to keep in mind: data alignment is intrinsic. The link between labels and data will not be broken unless done so explicitly by you.
We’ll give a brief intro to the data structures, then consider all of the broad categories of functionality and methods in separate sections.
When using pandas, we recommend the following import convention:
import pandas as pd
Series¶
Series is a one-dimensional labeled array (technically a subclass of ndarray) capable of holding any data type (integers, strings, floating point numbers, Python objects, etc.). The axis labels are collectively referred to as the index. The basic method to create a Series is to call:
>>> s = Series(data, index=index)
Here, data can be many different things:
- a Python dict
- an ndarray
- a scalar value (like 5)
The passed index is a list of axis labels. Thus, this separates into a few cases depending on what data is:
From ndarray
If data is an ndarray, index must be the same length as data. If no index is passed, one will be created having values [0, ..., len(data) - 1].
In [454]: s = Series(randn(5), index=['a', 'b', 'c', 'd', 'e'])
In [455]: s
Out[455]:
a -1.344
b 0.845
c 1.076
d -0.109
e 1.644
dtype: float64
In [456]: s.index
Out[456]: Index([a, b, c, d, e], dtype=object)
In [457]: Series(randn(5))
Out[457]:
0 -1.469
1 0.357
2 -0.675
3 -1.777
4 -0.969
dtype: float64
Note
Starting in v0.8.0, pandas supports non-unique index values. If an operation that does not support duplicate index values is attempted, an exception will be raised at that time. The reason for being lazy is nearly all performance-based (there are many instances in computations, like parts of GroupBy, where the index is not used).
From dict
If data is a dict, if index is passed the values in data corresponding to the labels in the index will be pulled out. Otherwise, an index will be constructed from the sorted keys of the dict, if possible.
In [458]: d = {'a' : 0., 'b' : 1., 'c' : 2.}
In [459]: Series(d)
Out[459]:
a 0
b 1
c 2
dtype: float64
In [460]: Series(d, index=['b', 'c', 'd', 'a'])
Out[460]:
b 1
c 2
d NaN
a 0
dtype: float64
Note
NaN (not a number) is the standard missing data marker used in pandas
From scalar value If data is a scalar value, an index must be provided. The value will be repeated to match the length of index
In [461]: Series(5., index=['a', 'b', 'c', 'd', 'e'])
Out[461]:
a 5
b 5
c 5
d 5
e 5
dtype: float64
Series is ndarray-like¶
As a subclass of ndarray, Series is a valid argument to most NumPy functions and behaves similarly to a NumPy array. However, things like slicing also slice the index.
In [462]: s[0]
Out[462]: -1.3443118127316671
In [463]: s[:3]
Out[463]:
a -1.344
b 0.845
c 1.076
dtype: float64
In [464]: s[s > s.median()]
Out[464]:
c 1.076
e 1.644
dtype: float64
In [465]: s[[4, 3, 1]]
Out[465]:
e 1.644
d -0.109
b 0.845
dtype: float64
In [466]: np.exp(s)
Out[466]:
a 0.261
b 2.328
c 2.932
d 0.897
e 5.174
dtype: float64
We will address array-based indexing in a separate section.
Series is dict-like¶
A Series is like a fixed-size dict in that you can get and set values by index label:
In [467]: s['a']
Out[467]: -1.3443118127316671
In [468]: s['e'] = 12.
In [469]: s
Out[469]:
a -1.344
b 0.845
c 1.076
d -0.109
e 12.000
dtype: float64
In [470]: 'e' in s
Out[470]: True
In [471]: 'f' in s
Out[471]: False
If a label is not contained, an exception is raised:
>>> s['f']
KeyError: 'f'
Using the get method, a missing label will return None or specified default:
In [472]: s.get('f')
In [473]: s.get('f', np.nan)
Out[473]: nan
Vectorized operations and label alignment with Series¶
When doing data analysis, as with raw NumPy arrays looping through Series value-by-value is usually not necessary. Series can be also be passed into most NumPy methods expecting an ndarray.
In [474]: s + s
Out[474]:
a -2.689
b 1.690
c 2.152
d -0.218
e 24.000
dtype: float64
In [475]: s * 2
Out[475]:
a -2.689
b 1.690
c 2.152
d -0.218
e 24.000
dtype: float64
In [476]: np.exp(s)
Out[476]:
a 0.261
b 2.328
c 2.932
d 0.897
e 162754.791
dtype: float64
A key difference between Series and ndarray is that operations between Series automatically align the data based on label. Thus, you can write computations without giving consideration to whether the Series involved have the same labels.
In [477]: s[1:] + s[:-1]
Out[477]:
a NaN
b 1.690
c 2.152
d -0.218
e NaN
dtype: float64
The result of an operation between unaligned Series will have the union of the indexes involved. If a label is not found in one Series or the other, the result will be marked as missing (NaN). Being able to write code without doing any explicit data alignment grants immense freedom and flexibility in interactive data analysis and research. The integrated data alignment features of the pandas data structures set pandas apart from the majority of related tools for working with labeled data.
Note
In general, we chose to make the default result of operations between differently indexed objects yield the union of the indexes in order to avoid loss of information. Having an index label, though the data is missing, is typically important information as part of a computation. You of course have the option of dropping labels with missing data via the dropna function.
Name attribute¶
Series can also have a name attribute:
In [478]: s = Series(np.random.randn(5), name='something')
In [479]: s
Out[479]:
0 -1.295
1 0.414
2 0.277
3 -0.472
4 -0.014
Name: something, dtype: float64
In [480]: s.name
Out[480]: 'something'
The Series name will be assigned automatically in many cases, in particular when taking 1D slices of DataFrame as you will see below.
DataFrame¶
DataFrame is a 2-dimensional labeled data structure with columns of potentially different types. You can think of it like a spreadsheet or SQL table, or a dict of Series objects. It is generally the most commonly used pandas object. Like Series, DataFrame accepts many different kinds of input:
- Dict of 1D ndarrays, lists, dicts, or Series
- 2-D numpy.ndarray
- Structured or record ndarray
- A Series
- Another DataFrame
Along with the data, you can optionally pass index (row labels) and columns (column labels) arguments. If you pass an index and / or columns, you are guaranteeing the index and / or columns of the resulting DataFrame. Thus, a dict of Series plus a specific index will discard all data not matching up to the passed index.
If axis labels are not passed, they will be constructed from the input data based on common sense rules.
From dict of Series or dicts¶
The result index will be the union of the indexes of the various Series. If there are any nested dicts, these will be first converted to Series. If no columns are passed, the columns will be the sorted list of dict keys.
In [481]: d = {'one' : Series([1., 2., 3.], index=['a', 'b', 'c']),
.....: 'two' : Series([1., 2., 3., 4.], index=['a', 'b', 'c', 'd'])}
.....:
In [482]: df = DataFrame(d)
In [483]: df
Out[483]:
one two
a 1 1
b 2 2
c 3 3
d NaN 4
In [484]: DataFrame(d, index=['d', 'b', 'a'])
Out[484]:
one two
d NaN 4
b 2 2
a 1 1
In [485]: DataFrame(d, index=['d', 'b', 'a'], columns=['two', 'three'])
Out[485]:
two three
d 4 NaN
b 2 NaN
a 1 NaN
The row and column labels can be accessed respectively by accessing the index and columns attributes:
Note
When a particular set of columns is passed along with a dict of data, the passed columns override the keys in the dict.
In [486]: df.index
Out[486]: Index([a, b, c, d], dtype=object)
In [487]: df.columns
Out[487]: Index([one, two], dtype=object)
From dict of ndarrays / lists¶
The ndarrays must all be the same length. If an index is passed, it must clearly also be the same length as the arrays. If no index is passed, the result will be range(n), where n is the array length.
In [488]: d = {'one' : [1., 2., 3., 4.],
.....: 'two' : [4., 3., 2., 1.]}
.....:
In [489]: DataFrame(d)
Out[489]:
one two
0 1 4
1 2 3
2 3 2
3 4 1
In [490]: DataFrame(d, index=['a', 'b', 'c', 'd'])
Out[490]:
one two
a 1 4
b 2 3
c 3 2
d 4 1
From structured or record array¶
This case is handled identically to a dict of arrays.
In [491]: data = np.zeros((2,),dtype=[('A', 'i4'),('B', 'f4'),('C', 'a10')])
In [492]: data[:] = [(1,2.,'Hello'),(2,3.,"World")]
In [493]: DataFrame(data)
Out[493]:
A B C
0 1 2 Hello
1 2 3 World
In [494]: DataFrame(data, index=['first', 'second'])
Out[494]:
A B C
first 1 2 Hello
second 2 3 World
In [495]: DataFrame(data, columns=['C', 'A', 'B'])
Out[495]:
C A B
0 Hello 1 2
1 World 2 3
Note
DataFrame is not intended to work exactly like a 2-dimensional NumPy ndarray.
From a list of dicts¶
In [496]: data2 = [{'a': 1, 'b': 2}, {'a': 5, 'b': 10, 'c': 20}]
In [497]: DataFrame(data2)
Out[497]:
a b c
0 1 2 NaN
1 5 10 20
In [498]: DataFrame(data2, index=['first', 'second'])
Out[498]:
a b c
first 1 2 NaN
second 5 10 20
In [499]: DataFrame(data2, columns=['a', 'b'])
Out[499]:
a b
0 1 2
1 5 10
From a Series¶
The result will be a DataFrame with the same index as the input Series, and with one column whose name is the original name of the Series (only if no other column name provided).
Missing Data
Much more will be said on this topic in the Missing data section. To construct a DataFrame with missing data, use np.nan for those values which are missing. Alternatively, you may pass a numpy.MaskedArray as the data argument to the DataFrame constructor, and its masked entries will be considered missing.
Alternate Constructors¶
DataFrame.from_dict
DataFrame.from_dict takes a dict of dicts or a dict of array-like sequences and returns a DataFrame. It operates like the DataFrame constructor except for the orient parameter which is 'columns' by default, but which can be set to 'index' in order to use the dict keys as row labels.
DataFrame.from_records
DataFrame.from_records takes a list of tuples or an ndarray with structured dtype. Works analogously to the normal DataFrame constructor, except that index maybe be a specific field of the structured dtype to use as the index. For example:
In [500]: data
Out[500]:
array([(1, 2.0, 'Hello'), (2, 3.0, 'World')],
dtype=[('A', '<i4'), ('B', '<f4'), ('C', 'S10')])
In [501]: DataFrame.from_records(data, index='C')
Out[501]:
A B
C
Hello 1 2
World 2 3
DataFrame.from_items
DataFrame.from_items works analogously to the form of the dict constructor that takes a sequence of (key, value) pairs, where the keys are column (or row, in the case of orient='index') names, and the value are the column values (or row values). This can be useful for constructing a DataFrame with the columns in a particular order without having to pass an explicit list of columns:
In [502]: DataFrame.from_items([('A', [1, 2, 3]), ('B', [4, 5, 6])])
Out[502]:
A B
0 1 4
1 2 5
2 3 6
If you pass orient='index', the keys will be the row labels. But in this case you must also pass the desired column names:
In [503]: DataFrame.from_items([('A', [1, 2, 3]), ('B', [4, 5, 6])],
.....: orient='index', columns=['one', 'two', 'three'])
.....:
Out[503]:
one two three
A 1 2 3
B 4 5 6
Column selection, addition, deletion¶
You can treat a DataFrame semantically like a dict of like-indexed Series objects. Getting, setting, and deleting columns works with the same syntax as the analogous dict operations:
In [504]: df['one']
Out[504]:
a 1
b 2
c 3
d NaN
Name: one, dtype: float64
In [505]: df['three'] = df['one'] * df['two']
In [506]: df['flag'] = df['one'] > 2
In [507]: df
Out[507]:
one two three flag
a 1 1 1 False
b 2 2 4 False
c 3 3 9 True
d NaN 4 NaN False
Columns can be deleted or popped like with a dict:
In [508]: del df['two']
In [509]: three = df.pop('three')
In [510]: df
Out[510]:
one flag
a 1 False
b 2 False
c 3 True
d NaN False
When inserting a scalar value, it will naturally be propagated to fill the column:
In [511]: df['foo'] = 'bar'
In [512]: df
Out[512]:
one flag foo
a 1 False bar
b 2 False bar
c 3 True bar
d NaN False bar
When inserting a Series that does not have the same index as the DataFrame, it will be conformed to the DataFrame’s index:
In [513]: df['one_trunc'] = df['one'][:2]
In [514]: df
Out[514]:
one flag foo one_trunc
a 1 False bar 1
b 2 False bar 2
c 3 True bar NaN
d NaN False bar NaN
You can insert raw ndarrays but their length must match the length of the DataFrame’s index.
By default, columns get inserted at the end. The insert function is available to insert at a particular location in the columns:
In [515]: df.insert(1, 'bar', df['one'])
In [516]: df
Out[516]:
one bar flag foo one_trunc
a 1 1 False bar 1
b 2 2 False bar 2
c 3 3 True bar NaN
d NaN NaN False bar NaN
Indexing / Selection¶
The basics of indexing are as follows:
Operation | Syntax | Result |
---|---|---|
Select column | df[col] | Series |
Select row by label | df.loc[label] | Series |
Select row by integer location | df.iloc[loc] | Series |
Slice rows | df[5:10] | DataFrame |
Select rows by boolean vector | df[bool_vec] | DataFrame |
Row selection, for example, returns a Series whose index is the columns of the DataFrame:
In [517]: df.loc['b']
Out[517]:
one 2
bar 2
flag False
foo bar
one_trunc 2
Name: b, dtype: object
In [518]: df.iloc[2]
Out[518]:
one 3
bar 3
flag True
foo bar
one_trunc NaN
Name: c, dtype: object
For a more exhaustive treatment of more sophisticated label-based indexing and slicing, see the section on indexing. We will address the fundamentals of reindexing / conforming to new sets of lables in the section on reindexing.
Data alignment and arithmetic¶
Data alignment between DataFrame objects automatically align on both the columns and the index (row labels). Again, the resulting object will have the union of the column and row labels.
In [519]: df = DataFrame(randn(10, 4), columns=['A', 'B', 'C', 'D'])
In [520]: df2 = DataFrame(randn(7, 3), columns=['A', 'B', 'C'])
In [521]: df + df2
Out[521]:
A B C D
0 -1.473 -0.626 -0.773 NaN
1 0.073 -0.519 2.742 NaN
2 1.744 -1.325 0.075 NaN
3 -1.366 -1.238 -1.782 NaN
4 0.275 -0.613 -2.263 NaN
5 1.263 2.338 1.260 NaN
6 -1.216 3.371 -1.992 NaN
7 NaN NaN NaN NaN
8 NaN NaN NaN NaN
9 NaN NaN NaN NaN
When doing an operation between DataFrame and Series, the default behavior is to align the Series index on the DataFrame columns, thus broadcasting row-wise. For example:
In [522]: df - df.iloc[0]
Out[522]:
A B C D
0 0.000 0.000 0.000 0.000
1 1.168 -1.200 3.489 0.536
2 1.703 -1.164 0.697 -0.485
3 1.176 0.138 0.096 -0.972
4 -0.825 1.136 -0.514 -2.309
5 1.970 1.030 1.493 -0.020
6 -1.849 0.981 -1.084 -1.306
7 0.284 0.552 -0.296 -2.123
8 1.132 -1.275 0.195 -1.017
9 0.265 0.702 1.265 0.064
In the special case of working with time series data, if the Series is a TimeSeries (which it will be automatically if the index contains datetime objects), and the DataFrame index also contains dates, the broadcasting will be column-wise:
In [523]: index = date_range('1/1/2000', periods=8)
In [524]: df = DataFrame(randn(8, 3), index=index,
.....: columns=['A', 'B', 'C'])
.....:
In [525]: df
Out[525]:
A B C
2000-01-01 3.357 -0.317 -1.236
2000-01-02 0.896 -0.488 -0.082
2000-01-03 -2.183 0.380 0.085
2000-01-04 0.432 1.520 -0.494
2000-01-05 0.600 0.274 0.133
2000-01-06 -0.024 2.410 1.451
2000-01-07 0.206 -0.252 -2.214
2000-01-08 1.063 1.266 0.299
In [526]: type(df['A'])
Out[526]: pandas.core.series.TimeSeries
In [527]: df - df['A']
Out[527]:
A B C
2000-01-01 0 -3.675 -4.594
2000-01-02 0 -1.384 -0.978
2000-01-03 0 2.563 2.268
2000-01-04 0 1.088 -0.926
2000-01-05 0 -0.326 -0.467
2000-01-06 0 2.434 1.474
2000-01-07 0 -0.458 -2.420
2000-01-08 0 0.203 -0.764
Technical purity aside, this case is so common in practice that supporting the special case is preferable to the alternative of forcing the user to transpose and do column-based alignment like so:
In [528]: (df.T - df['A']).T
Out[528]:
A B C
2000-01-01 0 -3.675 -4.594
2000-01-02 0 -1.384 -0.978
2000-01-03 0 2.563 2.268
2000-01-04 0 1.088 -0.926
2000-01-05 0 -0.326 -0.467
2000-01-06 0 2.434 1.474
2000-01-07 0 -0.458 -2.420
2000-01-08 0 0.203 -0.764
For explicit control over the matching and broadcasting behavior, see the section on flexible binary operations.
Operations with scalars are just as you would expect:
In [529]: df * 5 + 2
Out[529]:
A B C
2000-01-01 18.787 0.413 -4.181
2000-01-02 6.481 -0.438 1.589
2000-01-03 -8.915 3.902 2.424
2000-01-04 4.162 9.600 -0.468
2000-01-05 5.001 3.371 2.664
2000-01-06 1.882 14.051 9.253
2000-01-07 3.030 0.740 -9.068
2000-01-08 7.317 8.331 3.497
In [530]: 1 / df
Out[530]:
A B C
2000-01-01 0.298 -3.150 -0.809
2000-01-02 1.116 -2.051 -12.159
2000-01-03 -0.458 2.629 11.786
2000-01-04 2.313 0.658 -2.026
2000-01-05 1.666 3.647 7.525
2000-01-06 -42.215 0.415 0.689
2000-01-07 4.853 -3.970 -0.452
2000-01-08 0.940 0.790 3.340
In [531]: df ** 4
Out[531]:
A B C
2000-01-01 1.271e+02 0.010 2.336e+00
2000-01-02 6.450e-01 0.057 4.574e-05
2000-01-03 2.271e+01 0.021 5.182e-05
2000-01-04 3.495e-02 5.338 5.939e-02
2000-01-05 1.298e-01 0.006 3.118e-04
2000-01-06 3.149e-07 33.744 4.427e+00
2000-01-07 1.803e-03 0.004 2.401e+01
2000-01-08 1.278e+00 2.570 8.032e-03
Boolean operators work as well:
In [532]: df1 = DataFrame({'a' : [1, 0, 1], 'b' : [0, 1, 1] }, dtype=bool)
In [533]: df2 = DataFrame({'a' : [0, 1, 1], 'b' : [1, 1, 0] }, dtype=bool)
In [534]: df1 & df2
Out[534]:
a b
0 False False
1 False True
2 True False
In [535]: df1 | df2
Out[535]:
a b
0 True True
1 True True
2 True True
In [536]: df1 ^ df2
Out[536]:
a b
0 True True
1 True False
2 False True
In [537]: -df1
Out[537]:
a b
0 False True
1 True False
2 False False
Transposing¶
To transpose, access the T attribute (also the transpose function), similar to an ndarray:
# only show the first 5 rows
In [538]: df[:5].T
Out[538]:
2000-01-01 2000-01-02 2000-01-03 2000-01-04 2000-01-05
A 3.357 0.896 -2.183 0.432 0.600
B -0.317 -0.488 0.380 1.520 0.274
C -1.236 -0.082 0.085 -0.494 0.133
DataFrame interoperability with NumPy functions¶
Elementwise NumPy ufuncs (log, exp, sqrt, ...) and various other NumPy functions can be used with no issues on DataFrame, assuming the data within are numeric:
In [539]: np.exp(df)
Out[539]:
A B C
2000-01-01 28.715 0.728 0.290
2000-01-02 2.450 0.614 0.921
2000-01-03 0.113 1.463 1.089
2000-01-04 1.541 4.572 0.610
2000-01-05 1.822 1.316 1.142
2000-01-06 0.977 11.136 4.265
2000-01-07 1.229 0.777 0.109
2000-01-08 2.896 3.547 1.349
In [540]: np.asarray(df)
Out[540]:
array([[ 3.3574, -0.3174, -1.2363],
[ 0.8962, -0.4876, -0.0822],
[-2.1829, 0.3804, 0.0848],
[ 0.4324, 1.52 , -0.4937],
[ 0.6002, 0.2742, 0.1329],
[-0.0237, 2.4102, 1.4505],
[ 0.2061, -0.2519, -2.2136],
[ 1.0633, 1.2661, 0.2994]])
The dot method on DataFrame implements matrix multiplication:
In [541]: df.T.dot(df)
Out[541]:
A B C
A 18.562 -0.274 -4.715
B -0.274 10.344 4.184
C -4.715 4.184 8.897
Similarly, the dot method on Series implements dot product:
In [542]: s1 = Series(np.arange(5,10))
In [543]: s1.dot(s1)
Out[543]: 255
DataFrame is not intended to be a drop-in replacement for ndarray as its indexing semantics are quite different in places from a matrix.
Console display¶
For very large DataFrame objects, only a summary will be printed to the console (here I am reading a CSV version of the baseball dataset from the plyr R package):
In [544]: baseball = read_csv('data/baseball.csv')
In [545]: print baseball
<class 'pandas.core.frame.DataFrame'>
Int64Index: 100 entries, 88641 to 89534
Data columns (total 22 columns):
id 100 non-null values
year 100 non-null values
stint 100 non-null values
team 100 non-null values
lg 100 non-null values
g 100 non-null values
ab 100 non-null values
r 100 non-null values
h 100 non-null values
X2b 100 non-null values
X3b 100 non-null values
hr 100 non-null values
rbi 100 non-null values
sb 100 non-null values
cs 100 non-null values
bb 100 non-null values
so 100 non-null values
ibb 100 non-null values
hbp 100 non-null values
sh 100 non-null values
sf 100 non-null values
gidp 100 non-null values
dtypes: float64(9), int64(10), object(3)
However, using to_string will return a string representation of the DataFrame in tabular form, though it won’t always fit the console width:
In [546]: print baseball.iloc[-20:, :12].to_string()
id year stint team lg g ab r h X2b X3b hr
89474 finlest01 2007 1 COL NL 43 94 9 17 3 0 1
89480 embreal01 2007 1 OAK AL 4 0 0 0 0 0 0
89481 edmonji01 2007 1 SLN NL 117 365 39 92 15 2 12
89482 easleda01 2007 1 NYN NL 76 193 24 54 6 0 10
89489 delgaca01 2007 1 NYN NL 139 538 71 139 30 0 24
89493 cormirh01 2007 1 CIN NL 6 0 0 0 0 0 0
89494 coninje01 2007 2 NYN NL 21 41 2 8 2 0 0
89495 coninje01 2007 1 CIN NL 80 215 23 57 11 1 6
89497 clemero02 2007 1 NYA AL 2 2 0 1 0 0 0
89498 claytro01 2007 2 BOS AL 8 6 1 0 0 0 0
89499 claytro01 2007 1 TOR AL 69 189 23 48 14 0 1
89501 cirilje01 2007 2 ARI NL 28 40 6 8 4 0 0
89502 cirilje01 2007 1 MIN AL 50 153 18 40 9 2 2
89521 bondsba01 2007 1 SFN NL 126 340 75 94 14 0 28
89523 biggicr01 2007 1 HOU NL 141 517 68 130 31 3 10
89525 benitar01 2007 2 FLO NL 34 0 0 0 0 0 0
89526 benitar01 2007 1 SFN NL 19 0 0 0 0 0 0
89530 ausmubr01 2007 1 HOU NL 117 349 38 82 16 3 3
89533 aloumo01 2007 1 NYN NL 87 328 51 112 19 1 13
89534 alomasa02 2007 1 NYN NL 8 22 1 3 1 0 0
New since 0.10.0, wide DataFrames will now be printed across multiple rows by default:
In [547]: DataFrame(randn(3, 12))
Out[547]:
0 1 2 3 4 5 6 \
0 -0.863838 0.408204 -1.048089 -0.025747 -0.988387 0.094055 1.262731
1 0.369374 -0.034571 -2.484478 -0.281461 0.030711 0.109121 1.126203
2 -1.071357 0.441153 2.353925 0.583787 0.221471 -0.744471 0.758527
7 8 9 10 11
0 1.289997 0.082423 -0.055758 0.536580 -0.489682
1 -0.977349 1.474071 -0.064034 -1.282782 0.781836
2 1.729689 -0.964980 -0.845696 -1.340896 1.846883
You can change how much to print on a single row by setting the line_width option:
In [548]: set_option('line_width', 40) # default is 80
In [549]: DataFrame(randn(3, 12))
Out[549]:
0 1 2 \
0 -1.328865 1.682706 -1.717693
1 0.306996 -0.028665 0.384316
2 -1.137707 -0.891060 -0.693921
3 4 5 \
0 0.888782 0.228440 0.901805
1 1.574159 1.588931 0.476720
2 1.613616 0.464000 0.227371
6 7 8 \
0 1.171216 0.520260 -1.197071
1 0.473424 -0.242861 -0.014805
2 -0.496922 0.306389 -2.290613
9 10 11
0 -1.066969 -0.303421 -0.858447
1 -0.284319 0.650776 -1.461665
2 -1.134623 -1.561819 -0.260838
You can also disable this feature via the expand_frame_repr option:
In [550]: set_option('expand_frame_repr', False)
In [551]: DataFrame(randn(3, 12))
Out[551]:
<class 'pandas.core.frame.DataFrame'>
Int64Index: 3 entries, 0 to 2
Data columns (total 12 columns):
0 3 non-null values
1 3 non-null values
2 3 non-null values
3 3 non-null values
4 3 non-null values
5 3 non-null values
6 3 non-null values
7 3 non-null values
8 3 non-null values
9 3 non-null values
10 3 non-null values
11 3 non-null values
dtypes: float64(12)
DataFrame column attribute access and IPython completion¶
If a DataFrame column label is a valid Python variable name, the column can be accessed like attributes:
In [552]: df = DataFrame({'foo1' : np.random.randn(5),
.....: 'foo2' : np.random.randn(5)})
.....:
In [553]: df
Out[553]:
foo1 foo2
0 0.967661 -0.681087
1 -1.057909 0.377953
2 1.375020 0.493672
3 -0.928797 -2.461467
4 -0.308853 -1.553902
In [554]: df.foo1
Out[554]:
0 0.967661
1 -1.057909
2 1.375020
3 -0.928797
4 -0.308853
Name: foo1, dtype: float64
The columns are also connected to the IPython completion mechanism so they can be tab-completed:
In [5]: df.fo<TAB>
df.foo1 df.foo2
Panel¶
Panel is a somewhat less-used, but still important container for 3-dimensional data. The term panel data is derived from econometrics and is partially responsible for the name pandas: pan(el)-da(ta)-s. The names for the 3 axes are intended to give some semantic meaning to describing operations involving panel data and, in particular, econometric analysis of panel data. However, for the strict purposes of slicing and dicing a collection of DataFrame objects, you may find the axis names slightly arbitrary:
- items: axis 0, each item corresponds to a DataFrame contained inside
- major_axis: axis 1, it is the index (rows) of each of the DataFrames
- minor_axis: axis 2, it is the columns of each of the DataFrames
Construction of Panels works about like you would expect:
From 3D ndarray with optional axis labels¶
In [555]: wp = Panel(randn(2, 5, 4), items=['Item1', 'Item2'],
.....: major_axis=date_range('1/1/2000', periods=5),
.....: minor_axis=['A', 'B', 'C', 'D'])
.....:
In [556]: wp
Out[556]:
<class 'pandas.core.panel.Panel'>
Dimensions: 2 (items) x 5 (major_axis) x 4 (minor_axis)
Items axis: Item1 to Item2
Major_axis axis: 2000-01-01 00:00:00 to 2000-01-05 00:00:00
Minor_axis axis: A to D
From dict of DataFrame objects¶
In [557]: data = {'Item1' : DataFrame(randn(4, 3)),
.....: 'Item2' : DataFrame(randn(4, 2))}
.....:
In [558]: Panel(data)
Out[558]:
<class 'pandas.core.panel.Panel'>
Dimensions: 2 (items) x 4 (major_axis) x 3 (minor_axis)
Items axis: Item1 to Item2
Major_axis axis: 0 to 3
Minor_axis axis: 0 to 2
Note that the values in the dict need only be convertible to DataFrame. Thus, they can be any of the other valid inputs to DataFrame as per above.
One helpful factory method is Panel.from_dict, which takes a dictionary of DataFrames as above, and the following named parameters:
Parameter | Default | Description |
---|---|---|
intersect | False | drops elements whose indices do not align |
orient | items | use minor to use DataFrames’ columns as panel items |
For example, compare to the construction above:
In [559]: Panel.from_dict(data, orient='minor')
Out[559]:
<class 'pandas.core.panel.Panel'>
Dimensions: 3 (items) x 4 (major_axis) x 2 (minor_axis)
Items axis: 0 to 2
Major_axis axis: 0 to 3
Minor_axis axis: Item1 to Item2
Orient is especially useful for mixed-type DataFrames. If you pass a dict of DataFrame objects with mixed-type columns, all of the data will get upcasted to dtype=object unless you pass orient='minor':
In [560]: df = DataFrame({'a': ['foo', 'bar', 'baz'],
.....: 'b': np.random.randn(3)})
.....:
In [561]: df
Out[561]:
a b
0 foo -1.004168
1 bar -1.377627
2 baz 0.499281
In [562]: data = {'item1': df, 'item2': df}
In [563]: panel = Panel.from_dict(data, orient='minor')
In [564]: panel['a']
Out[564]:
item1 item2
0 foo foo
1 bar bar
2 baz baz
In [565]: panel['b']
Out[565]:
item1 item2
0 -1.004168 -1.004168
1 -1.377627 -1.377627
2 0.499281 0.499281
In [566]: panel['b'].dtypes
Out[566]:
item1 float64
item2 float64
dtype: object
Note
Unfortunately Panel, being less commonly used than Series and DataFrame, has been slightly neglected feature-wise. A number of methods and options available in DataFrame are not available in Panel. This will get worked on, of course, in future releases. And faster if you join me in working on the codebase.
From DataFrame using to_panel method¶
This method was introduced in v0.7 to replace LongPanel.to_long, and converts a DataFrame with a two-level index to a Panel.
In [567]: midx = MultiIndex(levels=[['one', 'two'], ['x','y']], labels=[[1,1,0,0],[1,0,1,0]])
In [568]: df = DataFrame({'A' : [1, 2, 3, 4], 'B': [5, 6, 7, 8]}, index=midx)
In [569]: df.to_panel()
Out[569]:
<class 'pandas.core.panel.Panel'>
Dimensions: 2 (items) x 2 (major_axis) x 2 (minor_axis)
Items axis: A to B
Major_axis axis: one to two
Minor_axis axis: x to y
Item selection / addition / deletion¶
Similar to DataFrame functioning as a dict of Series, Panel is like a dict of DataFrames:
In [570]: wp['Item1']
Out[570]:
A B C D
2000-01-01 2.015523 -1.833722 1.771740 -0.670027
2000-01-02 0.049307 -0.521493 -3.201750 0.792716
2000-01-03 0.146111 1.903247 -0.747169 -0.309038
2000-01-04 0.393876 1.861468 0.936527 1.255746
2000-01-05 -2.655452 1.219492 0.062297 -0.110388
In [571]: wp['Item3'] = wp['Item1'] / wp['Item2']
The API for insertion and deletion is the same as for DataFrame. And as with DataFrame, if the item is a valid python identifier, you can access it as an attribute and tab-complete it in IPython.
Transposing¶
A Panel can be rearranged using its transpose method (which does not make a copy by default unless the data are heterogeneous):
In [572]: wp.transpose(2, 0, 1)
Out[572]:
<class 'pandas.core.panel.Panel'>
Dimensions: 4 (items) x 3 (major_axis) x 5 (minor_axis)
Items axis: A to D
Major_axis axis: Item1 to Item3
Minor_axis axis: 2000-01-01 00:00:00 to 2000-01-05 00:00:00
Indexing / Selection¶
Operation | Syntax | Result |
---|---|---|
Select item | wp[item] | DataFrame |
Get slice at major_axis label | wp.major_xs(val) | DataFrame |
Get slice at minor_axis label | wp.minor_xs(val) | DataFrame |
For example, using the earlier example data, we could do:
In [573]: wp['Item1']
Out[573]:
A B C D
2000-01-01 2.015523 -1.833722 1.771740 -0.670027
2000-01-02 0.049307 -0.521493 -3.201750 0.792716
2000-01-03 0.146111 1.903247 -0.747169 -0.309038
2000-01-04 0.393876 1.861468 0.936527 1.255746
2000-01-05 -2.655452 1.219492 0.062297 -0.110388
In [574]: wp.major_xs(wp.major_axis[2])
Out[574]:
Item1 Item2 Item3
A 0.146111 -1.139050 -0.128275
B 1.903247 0.660342 2.882214
C -0.747169 0.464794 -1.607526
D -0.309038 -0.309337 0.999035
In [575]: wp.minor_axis
Out[575]: Index([A, B, C, D], dtype=object)
In [576]: wp.minor_xs('C')
Out[576]:
Item1 Item2 Item3
2000-01-01 1.771740 0.077849 22.758618
2000-01-02 -3.201750 0.503703 -6.356422
2000-01-03 -0.747169 0.464794 -1.607526
2000-01-04 0.936527 -0.643834 -1.454609
2000-01-05 0.062297 0.787872 0.079070
Squeezing¶
Another way to change the dimensionality of an object is to squeeze a 1-len object, similar to wp['Item1']
In [577]: wp.reindex(items=['Item1']).squeeze()
Out[577]:
A B C D
2000-01-01 2.015523 -1.833722 1.771740 -0.670027
2000-01-02 0.049307 -0.521493 -3.201750 0.792716
2000-01-03 0.146111 1.903247 -0.747169 -0.309038
2000-01-04 0.393876 1.861468 0.936527 1.255746
2000-01-05 -2.655452 1.219492 0.062297 -0.110388
In [578]: wp.reindex(items=['Item1'],minor=['B']).squeeze()
Out[578]:
2000-01-01 -1.833722
2000-01-02 -0.521493
2000-01-03 1.903247
2000-01-04 1.861468
2000-01-05 1.219492
Freq: D, Name: B, dtype: float64
Conversion to DataFrame¶
A Panel can be represented in 2D form as a hierarchically indexed DataFrame. See the section hierarchical indexing for more on this. To convert a Panel to a DataFrame, use the to_frame method:
In [579]: panel = Panel(np.random.randn(3, 5, 4), items=['one', 'two', 'three'],
.....: major_axis=date_range('1/1/2000', periods=5),
.....: minor_axis=['a', 'b', 'c', 'd'])
.....:
In [580]: panel.to_frame()
Out[580]:
one two three
major minor
2000-01-01 a -1.405256 -1.157886 0.086926
b 0.162565 -0.551865 -0.445645
c -0.067785 1.592673 -0.217503
d -1.260006 1.559318 -1.420361
2000-01-02 a -1.132896 1.562443 -0.015601
b -2.006481 0.763264 -1.150641
c 0.301016 0.162027 -0.798334
d 0.059117 -0.902704 -0.557697
2000-01-03 a 1.138469 1.106010 0.381353
b -2.400634 -0.199234 1.337122
c -0.280853 0.458265 -1.531095
d 0.025653 0.491048 1.331458
2000-01-04 a -1.386071 0.128594 -0.571329
b 0.863937 1.147862 -0.026671
c 0.252462 -1.256860 -1.085663
d 1.500571 0.563637 -1.114738
2000-01-05 a 1.053202 -2.417312 -0.058216
b -2.338595 0.972827 -0.486768
c -0.374279 0.041293 1.685148
d -2.359958 1.129659 0.112572
Panel4D (Experimental)¶
Panel4D is a 4-Dimensional named container very much like a Panel, but having 4 named dimensions. It is intended as a test bed for more N-Dimensional named containers.
- labels: axis 0, each item corresponds to a Panel contained inside
- items: axis 1, each item corresponds to a DataFrame contained inside
- major_axis: axis 2, it is the index (rows) of each of the DataFrames
- minor_axis: axis 3, it is the columns of each of the DataFrames
Panel4D is a sub-class of Panel, so most methods that work on Panels are applicable to Panel4D. The following methods are disabled:
- join , to_frame , to_excel , to_sparse , groupby
Construction of Panel4D works in a very similar manner to a Panel
From 4D ndarray with optional axis labels¶
In [581]: p4d = Panel4D(randn(2, 2, 5, 4),
.....: labels=['Label1','Label2'],
.....: items=['Item1', 'Item2'],
.....: major_axis=date_range('1/1/2000', periods=5),
.....: minor_axis=['A', 'B', 'C', 'D'])
.....:
In [582]: p4d
Out[582]:
<class 'pandas.core.panelnd.Panel4D'>
Dimensions: 2 (labels) x 2 (items) x 5 (major_axis) x 4 (minor_axis)
Labels axis: Label1 to Label2
Items axis: Item1 to Item2
Major_axis axis: 2000-01-01 00:00:00 to 2000-01-05 00:00:00
Minor_axis axis: A to D
From dict of Panel objects¶
In [583]: data = { 'Label1' : Panel({ 'Item1' : DataFrame(randn(4, 3)) }),
.....: 'Label2' : Panel({ 'Item2' : DataFrame(randn(4, 2)) }) }
.....:
In [584]: Panel4D(data)
Out[584]:
<class 'pandas.core.panelnd.Panel4D'>
Dimensions: 2 (labels) x 2 (items) x 4 (major_axis) x 3 (minor_axis)
Labels axis: Label1 to Label2
Items axis: Item1 to Item2
Major_axis axis: 0 to 3
Minor_axis axis: 0 to 2
Note that the values in the dict need only be convertible to Panels. Thus, they can be any of the other valid inputs to Panel as per above.
Slicing¶
Slicing works in a similar manner to a Panel. [] slices the first dimension. .ix allows you to slice abitrarily and get back lower dimensional objects
In [585]: p4d['Label1']
Out[585]:
<class 'pandas.core.panel.Panel'>
Dimensions: 2 (items) x 5 (major_axis) x 4 (minor_axis)
Items axis: Item1 to Item2
Major_axis axis: 2000-01-01 00:00:00 to 2000-01-05 00:00:00
Minor_axis axis: A to D
4D -> Panel
In [586]: p4d.ix[:,:,:,'A']
Out[586]:
<class 'pandas.core.panel.Panel'>
Dimensions: 2 (items) x 2 (major_axis) x 5 (minor_axis)
Items axis: Label1 to Label2
Major_axis axis: Item1 to Item2
Minor_axis axis: 2000-01-01 00:00:00 to 2000-01-05 00:00:00
4D -> DataFrame
In [587]: p4d.ix[:,:,0,'A']
Out[587]:
Label1 Label2
Item1 -1.495309 -0.739776
Item2 1.103949 0.403776
4D -> Series
In [588]: p4d.ix[:,0,0,'A']
Out[588]:
Label1 -1.495309
Label2 -0.739776
Name: A, dtype: float64
Transposing¶
A Panel4D can be rearranged using its transpose method (which does not make a copy by default unless the data are heterogeneous):
In [589]: p4d.transpose(3, 2, 1, 0)
Out[589]:
<class 'pandas.core.panelnd.Panel4D'>
Dimensions: 4 (labels) x 5 (items) x 2 (major_axis) x 2 (minor_axis)
Labels axis: A to D
Items axis: 2000-01-01 00:00:00 to 2000-01-05 00:00:00
Major_axis axis: Item1 to Item2
Minor_axis axis: Label1 to Label2
PanelND (Experimental)¶
PanelND is a module with a set of factory functions to enable a user to construct N-dimensional named containers like Panel4D, with a custom set of axis labels. Thus a domain-specific container can easily be created.
The following creates a Panel5D. A new panel type object must be sliceable into a lower dimensional object. Here we slice to a Panel4D.
In [590]: from pandas.core import panelnd
In [591]: Panel5D = panelnd.create_nd_panel_factory(
.....: klass_name = 'Panel5D',
.....: axis_orders = [ 'cool', 'labels','items','major_axis','minor_axis'],
.....: axis_slices = { 'labels' : 'labels', 'items' : 'items',
.....: 'major_axis' : 'major_axis', 'minor_axis' : 'minor_axis' },
.....: slicer = Panel4D,
.....: axis_aliases = { 'major' : 'major_axis', 'minor' : 'minor_axis' },
.....: stat_axis = 2)
.....:
In [592]: p5d = Panel5D(dict(C1 = p4d))
In [593]: p5d
Out[593]:
<class 'pandas.core.panelnd.Panel5D'>
Dimensions: 1 (cool) x 2 (labels) x 2 (items) x 5 (major_axis) x 4 (minor_axis)
Cool axis: C1 to C1
Labels axis: Label1 to Label2
Items axis: Item1 to Item2
Major_axis axis: 2000-01-01 00:00:00 to 2000-01-05 00:00:00
Minor_axis axis: A to D
# print a slice of our 5D
In [594]: p5d.ix['C1',:,:,0:3,:]
Out[594]:
<class 'pandas.core.panelnd.Panel4D'>
Dimensions: 2 (labels) x 2 (items) x 3 (major_axis) x 4 (minor_axis)
Labels axis: Label1 to Label2
Items axis: Item1 to Item2
Major_axis axis: 2000-01-01 00:00:00 to 2000-01-03 00:00:00
Minor_axis axis: A to D
# transpose it
In [595]: p5d.transpose(1,2,3,4,0)
Out[595]:
<class 'pandas.core.panelnd.Panel5D'>
Dimensions: 2 (cool) x 2 (labels) x 5 (items) x 4 (major_axis) x 1 (minor_axis)
Cool axis: Label1 to Label2
Labels axis: Item1 to Item2
Items axis: 2000-01-01 00:00:00 to 2000-01-05 00:00:00
Major_axis axis: A to D
Minor_axis axis: C1 to C1
# look at the shape & dim
In [596]: p5d.shape
Out[596]: [1, 2, 2, 5, 4]
In [597]: p5d.ndim
Out[597]: 5