Essential Basic Functionality — pandas 0.24.2 documentation (original) (raw)

Here we discuss a lot of the essential functionality common to the pandas data structures. Here’s how to create some of the objects used in the examples from the previous section:

In [1]: index = pd.date_range('1/1/2000', periods=8)

In [2]: s = pd.Series(np.random.randn(5), index=['a', 'b', 'c', 'd', 'e'])

In [3]: df = pd.DataFrame(np.random.randn(8, 3), index=index, ...: columns=['A', 'B', 'C']) ...:

In [4]: wp = pd.Panel(np.random.randn(2, 5, 4), items=['Item1', 'Item2'], ...: major_axis=pd.date_range('1/1/2000', periods=5), ...: minor_axis=['A', 'B', 'C', 'D']) ...:

Head and Tail

To view a small sample of a Series or DataFrame object, use thehead() and tail() methods. The default number of elements to display is five, but you may pass a custom number.

In [5]: long_series = pd.Series(np.random.randn(1000))

In [6]: long_series.head() Out[6]: 0 -2.211372 1 0.974466 2 -2.006747 3 -0.410001 4 -0.078638 dtype: float64

In [7]: long_series.tail(3) Out[7]: 997 -0.196166 998 0.380733 999 -0.275874 dtype: float64

Attributes and Underlying Data

pandas objects have a number of attributes enabling you to access the metadata

Note, these attributes can be safely assigned to!

In [8]: df[:2] Out[8]: A B C 2000-01-01 -0.173215 0.119209 -1.044236 2000-01-02 -0.861849 -2.104569 -0.494929

In [9]: df.columns = [x.lower() for x in df.columns]

In [10]: df Out[10]: a b c 2000-01-01 -0.173215 0.119209 -1.044236 2000-01-02 -0.861849 -2.104569 -0.494929 2000-01-03 1.071804 0.721555 -0.706771 2000-01-04 -1.039575 0.271860 -0.424972 2000-01-05 0.567020 0.276232 -1.087401 2000-01-06 -0.673690 0.113648 -1.478427 2000-01-07 0.524988 0.404705 0.577046 2000-01-08 -1.715002 -1.039268 -0.370647

Pandas objects (Index, Series, DataFrame) can be thought of as containers for arrays, which hold the actual data and do the actual computation. For many types, the underlying array is anumpy.ndarray. However, pandas and 3rd party libraries may _extend_NumPy’s type system to add support for custom arrays (see dtypes).

To get the actual data inside a Index or Series, use the .array property

In [11]: s.array Out[11]: [ 0.46911229990718628, -0.28286334432866328, -1.5090585031735124, -1.1356323710171934, 1.2121120250208506] Length: 5, dtype: float64

In [12]: s.index.array Out[12]: ['a', 'b', 'c', 'd', 'e'] Length: 5, dtype: object

array will always be an ExtensionArray. The exact details of what an ExtensionArray is and why pandas uses them is a bit beyond the scope of this introduction. See dtypes for more.

If you know you need a NumPy array, use to_numpy()or numpy.asarray().

In [13]: s.to_numpy() Out[13]: array([ 0.4691, -0.2829, -1.5091, -1.1356, 1.2121])

In [14]: np.asarray(s) Out[14]: array([ 0.4691, -0.2829, -1.5091, -1.1356, 1.2121])

When the Series or Index is backed by an ExtensionArray, to_numpy()may involve copying data and coercing values. See dtypes for more.

to_numpy() gives some control over the dtype of the resulting numpy.ndarray. For example, consider datetimes with timezones. NumPy doesn’t have a dtype to represent timezone-aware datetimes, so there are two possibly useful representations:

  1. An object-dtype numpy.ndarray with Timestamp objects, each with the correct tz
  2. A datetime64[ns] -dtype numpy.ndarray, where the values have been converted to UTC and the timezone discarded

Timezones may be preserved with dtype=object

In [15]: ser = pd.Series(pd.date_range('2000', periods=2, tz="CET"))

In [16]: ser.to_numpy(dtype=object) Out[16]: array([Timestamp('2000-01-01 00:00:00+0100', tz='CET', freq='D'), Timestamp('2000-01-02 00:00:00+0100', tz='CET', freq='D')], dtype=object)

Or thrown away with dtype='datetime64[ns]'

In [17]: ser.to_numpy(dtype="datetime64[ns]") Out[17]: array(['1999-12-31T23:00:00.000000000', '2000-01-01T23:00:00.000000000'], dtype='datetime64[ns]')

Getting the “raw data” inside a DataFrame is possibly a bit more complex. When your DataFrame only has a single data type for all the columns, DataFrame.to_numpy() will return the underlying data:

In [18]: df.to_numpy() Out[18]: array([[-0.1732, 0.1192, -1.0442], [-0.8618, -2.1046, -0.4949], [ 1.0718, 0.7216, -0.7068], [-1.0396, 0.2719, -0.425 ], [ 0.567 , 0.2762, -1.0874], [-0.6737, 0.1136, -1.4784], [ 0.525 , 0.4047, 0.577 ], [-1.715 , -1.0393, -0.3706]])

If a DataFrame or Panel contains homogeneously-typed data, the ndarray can actually be modified in-place, and the changes will be reflected in the data structure. For heterogeneous data (e.g. some of the DataFrame’s columns are not all the same dtype), this will not be the case. The values attribute itself, unlike the axis labels, cannot be assigned to.

Note

When working with heterogeneous data, the dtype of the resulting ndarray will be chosen to accommodate all of the data involved. For example, if strings are involved, the result will be of object dtype. If there are only floats and integers, the resulting array will be of float dtype.

In the past, pandas recommended Series.values or DataFrame.valuesfor extracting the data from a Series or DataFrame. You’ll still find references to these in old code bases and online. Going forward, we recommend avoiding.values and using .array or .to_numpy(). .values has the following drawbacks:

  1. When your Series contains an extension type, it’s unclear whether Series.values returns a NumPy array or the extension array.Series.array will always return an ExtensionArray, and will never copy data. Series.to_numpy() will always return a NumPy array, potentially at the cost of copying / coercing values.
  2. When your DataFrame contains a mixture of data types, DataFrame.values may involve copying data and coercing values to a common dtype, a relatively expensive operation. DataFrame.to_numpy(), being a method, makes it clearer that the returned NumPy array may not be a view on the same data in the DataFrame.

Accelerated operations

pandas has support for accelerating certain types of binary numerical and boolean operations using the numexpr library and the bottleneck libraries.

These libraries are especially useful when dealing with large data sets, and provide large speedups. numexpr uses smart chunking, caching, and multiple cores. bottleneck is a set of specialized cython routines that are especially fast when dealing with arrays that havenans.

Here is a sample (using 100 column x 100,000 row DataFrames):

Operation 0.11.0 (ms) Prior Version (ms) Ratio to Prior
df1 > df2 13.32 125.35 0.1063
df1 * df2 21.71 36.63 0.5928
df1 + df2 22.04 36.50 0.6039

You are highly encouraged to install both libraries. See the sectionRecommended Dependencies for more installation info.

These are both enabled to be used by default, you can control this by setting the options:

New in version 0.20.0.

pd.set_option('compute.use_bottleneck', False) pd.set_option('compute.use_numexpr', False)

Flexible binary operations

With binary operations between pandas data structures, there are two key points of interest:

We will demonstrate how to manage these issues independently, though they can be handled simultaneously.

Matching / broadcasting behavior

DataFrame has the methods add(), sub(),mul(), div() and related functionsradd(), rsub(), … for carrying out binary operations. For broadcasting behavior, Series input is of primary interest. Using these functions, you can use to either match on the index or columns via the axis keyword:

In [19]: df = pd.DataFrame({ ....: 'one': pd.Series(np.random.randn(3), index=['a', 'b', 'c']), ....: 'two': pd.Series(np.random.randn(4), index=['a', 'b', 'c', 'd']), ....: 'three': pd.Series(np.random.randn(3), index=['b', 'c', 'd'])}) ....:

In [20]: df Out[20]: one two three a 1.400810 -1.643041 NaN b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 d NaN 1.553693 -1.670830

In [21]: row = df.iloc[1]

In [22]: column = df['two']

In [23]: df.sub(row, axis='columns') Out[23]: one two three a 1.757280 -2.688953 NaN b 0.000000 0.000000 0.000000 c 1.153738 -0.121396 -0.402113 d NaN 0.507782 -2.065853

In [24]: df.sub(row, axis=1) Out[24]: one two three a 1.757280 -2.688953 NaN b 0.000000 0.000000 0.000000 c 1.153738 -0.121396 -0.402113 d NaN 0.507782 -2.065853

In [25]: df.sub(column, axis='index') Out[25]: one two three a 3.043851 0.0 NaN b -1.402381 0.0 -0.650888 c -0.127247 0.0 -0.931605 d NaN 0.0 -3.224524

In [26]: df.sub(column, axis=0) Out[26]: one two three a 3.043851 0.0 NaN b -1.402381 0.0 -0.650888 c -0.127247 0.0 -0.931605 d NaN 0.0 -3.224524

Furthermore you can align a level of a MultiIndexed DataFrame with a Series.

In [27]: dfmi = df.copy()

In [28]: dfmi.index = pd.MultiIndex.from_tuples([(1, 'a'), (1, 'b'), ....: (1, 'c'), (2, 'a')], ....: names=['first', 'second']) ....:

In [29]: dfmi.sub(column, axis=0, level='second') Out[29]: one two three first second
1 a 3.043851 0.000000 NaN b -1.402381 0.000000 -0.650888 c -0.127247 0.000000 -0.931605 2 a NaN 3.196734 -0.027789

With Panel, describing the matching behavior is a bit more difficult, so the arithmetic methods instead (and perhaps confusingly?) give you the option to specify the broadcast axis. For example, suppose we wished to demean the data over a particular axis. This can be accomplished by taking the mean over an axis and broadcasting over the same axis:

In [30]: major_mean = wp.mean(axis='major')

In [31]: major_mean Out[31]: Item1 Item2 A -0.378069 0.675929 B -0.241429 -0.018080 C -0.597702 0.129006 D 0.204005 0.245570

In [32]: wp.sub(major_mean, axis='major') Out[32]: <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

And similarly for axis="items" and axis="minor".

Note

I could be convinced to make the axis argument in the DataFrame methods match the broadcasting behavior of Panel. Though it would require a transition period so users can change their code…

Series and Index also support the divmod() builtin. This function takes the floor division and modulo operation at the same time returning a two-tuple of the same type as the left hand side. For example:

In [33]: s = pd.Series(np.arange(10))

In [34]: s Out[34]: 0 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 dtype: int64

In [35]: div, rem = divmod(s, 3)

In [36]: div Out[36]: 0 0 1 0 2 0 3 1 4 1 5 1 6 2 7 2 8 2 9 3 dtype: int64

In [37]: rem Out[37]: 0 0 1 1 2 2 3 0 4 1 5 2 6 0 7 1 8 2 9 0 dtype: int64

In [38]: idx = pd.Index(np.arange(10))

In [39]: idx Out[39]: Int64Index([0, 1, 2, 3, 4, 5, 6, 7, 8, 9], dtype='int64')

In [40]: div, rem = divmod(idx, 3)

In [41]: div Out[41]: Int64Index([0, 0, 0, 1, 1, 1, 2, 2, 2, 3], dtype='int64')

In [42]: rem Out[42]: Int64Index([0, 1, 2, 0, 1, 2, 0, 1, 2, 0], dtype='int64')

We can also do elementwise divmod():

In [43]: div, rem = divmod(s, [2, 2, 3, 3, 4, 4, 5, 5, 6, 6])

In [44]: div Out[44]: 0 0 1 0 2 0 3 1 4 1 5 1 6 1 7 1 8 1 9 1 dtype: int64

In [45]: rem Out[45]: 0 0 1 1 2 2 3 0 4 0 5 1 6 1 7 2 8 2 9 3 dtype: int64

Missing data / operations with fill values

In Series and DataFrame, the arithmetic functions have the option of inputting a fill_value, namely a value to substitute when at most one of the values at a location are missing. For example, when adding two DataFrame objects, you may wish to treat NaN as 0 unless both DataFrames are missing that value, in which case the result will be NaN (you can later replace NaN with some other value using fillna if you wish).

In [46]: df Out[46]: one two three a 1.400810 -1.643041 NaN b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 d NaN 1.553693 -1.670830

In [47]: df2 Out[47]: one two three a 1.400810 -1.643041 1.000000 b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 d NaN 1.553693 -1.670830

In [48]: df + df2 Out[48]: one two three a 2.801620 -3.286083 NaN b -0.712940 2.091822 0.790046 c 1.594536 1.849030 -0.014180 d NaN 3.107386 -3.341661

In [49]: df.add(df2, fill_value=0) Out[49]: one two three a 2.801620 -3.286083 1.000000 b -0.712940 2.091822 0.790046 c 1.594536 1.849030 -0.014180 d NaN 3.107386 -3.341661

Flexible Comparisons

Series and DataFrame have the binary comparison methods eq, ne, lt, gt,le, and ge whose behavior is analogous to the binary arithmetic operations described above:

In [50]: df.gt(df2) Out[50]: one two three a False False False b False False False c False False False d False False False

In [51]: df2.ne(df) Out[51]: one two three a False False True b False False False c False False False d True False False

These operations produce a pandas object of the same type as the left-hand-side input that is of dtype bool. These boolean objects can be used in indexing operations, see the section on Boolean indexing.

Boolean Reductions

You can apply the reductions: empty, any(),all(), and bool() to provide a way to summarize a boolean result.

In [52]: (df > 0).all() Out[52]: one False two False three False dtype: bool

In [53]: (df > 0).any() Out[53]: one True two True three True dtype: bool

You can reduce to a final boolean value.

In [54]: (df > 0).any().any() Out[54]: True

You can test if a pandas object is empty, via the empty property.

In [55]: df.empty Out[55]: False

In [56]: pd.DataFrame(columns=list('ABC')).empty Out[56]: True

To evaluate single-element pandas objects in a boolean context, use the methodbool():

In [57]: pd.Series([True]).bool() Out[57]: True

In [58]: pd.Series([False]).bool() Out[58]: False

In [59]: pd.DataFrame([[True]]).bool() Out[59]: True

In [60]: pd.DataFrame([[False]]).bool() Out[60]: False

Warning

You might be tempted to do the following:

Or

These will both raise errors, as you are trying to compare multiple values.:

ValueError: The truth value of an array is ambiguous. Use a.empty, a.any() or a.all().

See gotchas for a more detailed discussion.

Comparing if objects are equivalent

Often you may find that there is more than one way to compute the same result. As a simple example, consider df + df and df * 2. To test that these two computations produce the same result, given the tools shown above, you might imagine using (df + df == df * 2).all(). But in fact, this expression is False:

In [61]: df + df == df * 2 Out[61]: one two three a True True False b True True True c True True True d False True True

In [62]: (df + df == df * 2).all() Out[62]: one False two True three False dtype: bool

Notice that the boolean DataFrame df + df == df * 2 contains some False values! This is because NaNs do not compare as equals:

In [63]: np.nan == np.nan Out[63]: False

So, NDFrames (such as Series, DataFrames, and Panels) have an equals() method for testing equality, with NaNs in corresponding locations treated as equal.

In [64]: (df + df).equals(df * 2) Out[64]: True

Note that the Series or DataFrame index needs to be in the same order for equality to be True:

In [65]: df1 = pd.DataFrame({'col': ['foo', 0, np.nan]})

In [66]: df2 = pd.DataFrame({'col': [np.nan, 0, 'foo']}, index=[2, 1, 0])

In [67]: df1.equals(df2) Out[67]: False

In [68]: df1.equals(df2.sort_index()) Out[68]: True

Comparing array-like objects

You can conveniently perform element-wise comparisons when comparing a pandas data structure with a scalar value:

In [69]: pd.Series(['foo', 'bar', 'baz']) == 'foo' Out[69]: 0 True 1 False 2 False dtype: bool

In [70]: pd.Index(['foo', 'bar', 'baz']) == 'foo' Out[70]: array([ True, False, False], dtype=bool)

Pandas also handles element-wise comparisons between different array-like objects of the same length:

In [71]: pd.Series(['foo', 'bar', 'baz']) == pd.Index(['foo', 'bar', 'qux']) Out[71]: 0 True 1 True 2 False dtype: bool

In [72]: pd.Series(['foo', 'bar', 'baz']) == np.array(['foo', 'bar', 'qux']) Out[72]: 0 True 1 True 2 False dtype: bool

Trying to compare Index or Series objects of different lengths will raise a ValueError:

In [55]: pd.Series(['foo', 'bar', 'baz']) == pd.Series(['foo', 'bar']) ValueError: Series lengths must match to compare

In [56]: pd.Series(['foo', 'bar', 'baz']) == pd.Series(['foo']) ValueError: Series lengths must match to compare

Note that this is different from the NumPy behavior where a comparison can be broadcast:

In [73]: np.array([1, 2, 3]) == np.array([2]) Out[73]: array([False, True, False], dtype=bool)

or it can return False if broadcasting can not be done:

In [74]: np.array([1, 2, 3]) == np.array([1, 2]) Out[74]: False

Combining overlapping data sets

A problem occasionally arising is the combination of two similar data sets where values in one are preferred over the other. An example would be two data series representing a particular economic indicator where one is considered to be of “higher quality”. However, the lower quality series might extend further back in history or have more complete data coverage. As such, we would like to combine two DataFrame objects where missing values in one DataFrame are conditionally filled with like-labeled values from the other DataFrame. The function implementing this operation is combine_first(), which we illustrate:

In [75]: df1 = pd.DataFrame({'A': [1., np.nan, 3., 5., np.nan], ....: 'B': [np.nan, 2., 3., np.nan, 6.]}) ....:

In [76]: df2 = pd.DataFrame({'A': [5., 2., 4., np.nan, 3., 7.], ....: 'B': [np.nan, np.nan, 3., 4., 6., 8.]}) ....:

In [77]: df1 Out[77]: A B 0 1.0 NaN 1 NaN 2.0 2 3.0 3.0 3 5.0 NaN 4 NaN 6.0

In [78]: df2 Out[78]: A B 0 5.0 NaN 1 2.0 NaN 2 4.0 3.0 3 NaN 4.0 4 3.0 6.0 5 7.0 8.0

In [79]: df1.combine_first(df2) Out[79]: A B 0 1.0 NaN 1 2.0 2.0 2 3.0 3.0 3 5.0 4.0 4 3.0 6.0 5 7.0 8.0

General DataFrame Combine

The combine_first() method above calls the more generalDataFrame.combine(). This method takes another DataFrame and a combiner function, aligns the input DataFrame and then passes the combiner function pairs of Series (i.e., columns whose names are the same).

So, for instance, to reproduce combine_first() as above:

In [80]: def combiner(x, y): ....: return np.where(pd.isna(x), y, x) ....:

Descriptive statistics

There exists a large number of methods for computing descriptive statistics and other related operations on Series, DataFrame, and Panel. Most of these are aggregations (hence producing a lower-dimensional result) likesum(), mean(), and quantile(), but some of them, like cumsum() and cumprod(), produce an object of the same size. Generally speaking, these methods take anaxis argument, just like ndarray.{sum, std, …}, but the axis can be specified by name or integer:

For example:

In [81]: df Out[81]: one two three a 1.400810 -1.643041 NaN b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 d NaN 1.553693 -1.670830

In [82]: df.mean(0) Out[82]: one 0.613869 two 0.470270 three -0.427633 dtype: float64

In [83]: df.mean(1) Out[83]: a -0.121116 b 0.361488 c 0.571564 d -0.058569 dtype: float64

All such methods have a skipna option signaling whether to exclude missing data (True by default):

In [84]: df.sum(0, skipna=False) Out[84]: one NaN two 1.881078 three NaN dtype: float64

In [85]: df.sum(axis=1, skipna=True) Out[85]: a -0.242232 b 1.084464 c 1.714693 d -0.117137 dtype: float64

Combined with the broadcasting / arithmetic behavior, one can describe various statistical procedures, like standardization (rendering data zero mean and standard deviation 1), very concisely:

In [86]: ts_stand = (df - df.mean()) / df.std()

In [87]: ts_stand.std() Out[87]: one 1.0 two 1.0 three 1.0 dtype: float64

In [88]: xs_stand = df.sub(df.mean(1), axis=0).div(df.std(1), axis=0)

In [89]: xs_stand.std(1) Out[89]: a 1.0 b 1.0 c 1.0 d 1.0 dtype: float64

Note that methods like cumsum() and cumprod()preserve the location of NaN values. This is somewhat different fromexpanding() and rolling(). For more details please see this note.

In [90]: df.cumsum() Out[90]: one two three a 1.400810 -1.643041 NaN b 1.044340 -0.597130 0.395023 c 1.841608 0.327385 0.387933 d NaN 1.881078 -1.282898

Here is a quick reference summary table of common functions. Each also takes an optional level parameter which applies only if the object has ahierarchical index.

Function Description
count Number of non-NA observations
sum Sum of values
mean Mean of values
mad Mean absolute deviation
median Arithmetic median of values
min Minimum
max Maximum
mode Mode
abs Absolute Value
prod Product of values
std Bessel-corrected sample standard deviation
var Unbiased variance
sem Standard error of the mean
skew Sample skewness (3rd moment)
kurt Sample kurtosis (4th moment)
quantile Sample quantile (value at %)
cumsum Cumulative sum
cumprod Cumulative product
cummax Cumulative maximum
cummin Cumulative minimum

Note that by chance some NumPy methods, like mean, std, and sum, will exclude NAs on Series input by default:

In [91]: np.mean(df['one']) Out[91]: 0.6138692844180106

In [92]: np.mean(df['one'].to_numpy()) Out[92]: nan

Series.nunique() will return the number of unique non-NA values in a Series:

In [93]: series = pd.Series(np.random.randn(500))

In [94]: series[20:500] = np.nan

In [95]: series[10:20] = 5

In [96]: series.nunique() Out[96]: 11

Summarizing data: describe

There is a convenient describe() function which computes a variety of summary statistics about a Series or the columns of a DataFrame (excluding NAs of course):

In [97]: series = pd.Series(np.random.randn(1000))

In [98]: series[::2] = np.nan

In [99]: series.describe() Out[99]: count 500.000000 mean -0.020695 std 1.011840 min -2.683763 25% -0.709297 50% -0.070211 75% 0.712856 max 3.160915 dtype: float64

In [100]: frame = pd.DataFrame(np.random.randn(1000, 5), .....: columns=['a', 'b', 'c', 'd', 'e']) .....:

In [101]: frame.iloc[::2] = np.nan

In [102]: frame.describe() Out[102]: a b c d e count 500.000000 500.000000 500.000000 500.000000 500.000000 mean 0.026515 0.022952 -0.047307 -0.052551 0.011210 std 1.016752 0.980046 1.020837 1.008271 1.006726 min -3.000951 -2.637901 -3.303099 -3.159200 -3.188821 25% -0.647623 -0.593587 -0.709906 -0.691338 -0.689176 50% 0.047578 -0.026675 -0.029655 -0.032769 -0.015775 75% 0.723946 0.771931 0.603753 0.667044 0.652221 max 2.740139 2.752332 3.004229 2.728702 3.240991

You can select specific percentiles to include in the output:

In [103]: series.describe(percentiles=[.05, .25, .75, .95]) Out[103]: count 500.000000 mean -0.020695 std 1.011840 min -2.683763 5% -1.641337 25% -0.709297 50% -0.070211 75% 0.712856 95% 1.699176 max 3.160915 dtype: float64

By default, the median is always included.

For a non-numerical Series object, describe() will give a simple summary of the number of unique values and most frequently occurring values:

In [104]: s = pd.Series(['a', 'a', 'b', 'b', 'a', 'a', np.nan, 'c', 'd', 'a'])

In [105]: s.describe() Out[105]: count 9 unique 4 top a freq 5 dtype: object

Note that on a mixed-type DataFrame object, describe() will restrict the summary to include only numerical columns or, if none are, only categorical columns:

In [106]: frame = pd.DataFrame({'a': ['Yes', 'Yes', 'No', 'No'], 'b': range(4)})

In [107]: frame.describe() Out[107]: b count 4.000000 mean 1.500000 std 1.290994 min 0.000000 25% 0.750000 50% 1.500000 75% 2.250000 max 3.000000

This behavior can be controlled by providing a list of types as include/excludearguments. The special value all can also be used:

In [108]: frame.describe(include=['object']) Out[108]: a count 4 unique 2 top Yes freq 2

In [109]: frame.describe(include=['number']) Out[109]: b count 4.000000 mean 1.500000 std 1.290994 min 0.000000 25% 0.750000 50% 1.500000 75% 2.250000 max 3.000000

In [110]: frame.describe(include='all') Out[110]: a b count 4 4.000000 unique 2 NaN top Yes NaN freq 2 NaN mean NaN 1.500000 std NaN 1.290994 min NaN 0.000000 25% NaN 0.750000 50% NaN 1.500000 75% NaN 2.250000 max NaN 3.000000

That feature relies on select_dtypes. Refer to there for details about accepted inputs.

Index of Min/Max Values

The idxmin() and idxmax() functions on Series and DataFrame compute the index labels with the minimum and maximum corresponding values:

In [111]: s1 = pd.Series(np.random.randn(5))

In [112]: s1 Out[112]: 0 -0.068822 1 -1.129788 2 -0.269798 3 -0.375580 4 0.513381 dtype: float64

In [113]: s1.idxmin(), s1.idxmax() Out[113]: (1, 4)

In [114]: df1 = pd.DataFrame(np.random.randn(5, 3), columns=['A', 'B', 'C'])

In [115]: df1 Out[115]: A B C 0 0.333329 -0.910090 -1.321220 1 2.111424 1.701169 0.858336 2 -0.608055 -2.082155 -0.069618 3 1.412817 -0.562658 0.770042 4 0.373294 -0.965381 -1.607840

In [116]: df1.idxmin(axis=0) Out[116]: A 2 B 2 C 4 dtype: int64

In [117]: df1.idxmax(axis=1) Out[117]: 0 A 1 A 2 C 3 A 4 A dtype: object

When there are multiple rows (or columns) matching the minimum or maximum value, idxmin() and idxmax() return the first matching index:

In [118]: df3 = pd.DataFrame([2, 1, 1, 3, np.nan], columns=['A'], index=list('edcba'))

In [119]: df3 Out[119]: A e 2.0 d 1.0 c 1.0 b 3.0 a NaN

In [120]: df3['A'].idxmin() Out[120]: 'd'

Note

idxmin and idxmax are called argmin and argmax in NumPy.

Value counts (histogramming) / Mode

The value_counts() Series method and top-level function computes a histogram of a 1D array of values. It can also be used as a function on regular arrays:

In [121]: data = np.random.randint(0, 7, size=50)

In [122]: data Out[122]: array([6, 4, 1, 3, 4, 4, 4, 6, 5, 2, 6, 1, 0, 4, 3, 2, 5, 3, 4, 0, 5, 3, 0, 1, 5, 0, 1, 5, 3, 4, 1, 2, 3, 2, 4, 6, 1, 4, 3, 5, 2, 1, 2, 4, 1, 6, 3, 6, 3, 3])

In [123]: s = pd.Series(data)

In [124]: s.value_counts() Out[124]: 4 10 3 10 1 8 6 6 5 6 2 6 0 4 dtype: int64

In [125]: pd.value_counts(data) Out[125]: 4 10 3 10 1 8 6 6 5 6 2 6 0 4 dtype: int64

Similarly, you can get the most frequently occurring value(s) (the mode) of the values in a Series or DataFrame:

In [126]: s5 = pd.Series([1, 1, 3, 3, 3, 5, 5, 7, 7, 7])

In [127]: s5.mode() Out[127]: 0 3 1 7 dtype: int64

In [128]: df5 = pd.DataFrame({"A": np.random.randint(0, 7, size=50), .....: "B": np.random.randint(-10, 15, size=50)}) .....:

In [129]: df5.mode() Out[129]: A B 0 0 -9

Discretization and quantiling

Continuous values can be discretized using the cut() (bins based on values) and qcut() (bins based on sample quantiles) functions:

In [130]: arr = np.random.randn(20)

In [131]: factor = pd.cut(arr, 4)

In [132]: factor Out[132]: [(1.27, 2.31], (0.231, 1.27], (-0.809, 0.231], (-1.853, -0.809], (1.27, 2.31], ..., (0.231, 1.27], (-0.809, 0.231], (-1.853, -0.809], (1.27, 2.31], (0.231, 1.27]] Length: 20 Categories (4, interval[float64]): [(-1.853, -0.809] < (-0.809, 0.231] < (0.231, 1.27] < (1.27, 2.31]]

In [133]: factor = pd.cut(arr, [-5, -1, 0, 1, 5])

In [134]: factor Out[134]: [(1, 5], (0, 1], (-1, 0], (-5, -1], (1, 5], ..., (1, 5], (-1, 0], (-5, -1], (1, 5], (0, 1]] Length: 20 Categories (4, interval[int64]): [(-5, -1] < (-1, 0] < (0, 1] < (1, 5]]

qcut() computes sample quantiles. For example, we could slice up some normally distributed data into equal-size quartiles like so:

In [135]: arr = np.random.randn(30)

In [136]: factor = pd.qcut(arr, [0, .25, .5, .75, 1])

In [137]: factor Out[137]: [(-2.219, -0.669], (-0.669, 0.00453], (0.367, 2.369], (0.00453, 0.367], (0.367, 2.369], ..., (0.00453, 0.367], (0.367, 2.369], (0.00453, 0.367], (-0.669, 0.00453], (0.367, 2.369]] Length: 30 Categories (4, interval[float64]): [(-2.219, -0.669] < (-0.669, 0.00453] < (0.00453, 0.367] < (0.367, 2.369]]

In [138]: pd.value_counts(factor) Out[138]: (0.367, 2.369] 8 (-2.219, -0.669] 8 (0.00453, 0.367] 7 (-0.669, 0.00453] 7 dtype: int64

We can also pass infinite values to define the bins:

In [139]: arr = np.random.randn(20)

In [140]: factor = pd.cut(arr, [-np.inf, 0, np.inf])

In [141]: factor Out[141]: [(0.0, inf], (-inf, 0.0], (-inf, 0.0], (-inf, 0.0], (-inf, 0.0], ..., (-inf, 0.0], (-inf, 0.0], (0.0, inf], (-inf, 0.0], (-inf, 0.0]] Length: 20 Categories (2, interval[float64]): [(-inf, 0.0] < (0.0, inf]]

Function application

To apply your own or another library’s functions to pandas objects, you should be aware of the three methods below. The appropriate method to use depends on whether your function expects to operate on an entire DataFrame or Series, row- or column-wise, or elementwise.

  1. Tablewise Function Application: pipe()
  2. Row or Column-wise Function Application: apply()
  3. Aggregation API: agg() and transform()
  4. Applying Elementwise Functions: applymap()

Tablewise Function Application

DataFrames and Series can of course just be passed into functions. However, if the function needs to be called in a chain, consider using the pipe() method. Compare the following

f, g, and h are functions taking and returning DataFrames

f(g(h(df), arg1=1), arg2=2, arg3=3)

with the equivalent

(df.pipe(h) ... .pipe(g, arg1=1) ... .pipe(f, arg2=2, arg3=3))

Pandas encourages the second style, which is known as method chaining.pipe makes it easy to use your own or another library’s functions in method chains, alongside pandas’ methods.

In the example above, the functions f, g, and h each expected the DataFrame as the first positional argument. What if the function you wish to apply takes its data as, say, the second argument? In this case, provide pipe with a tuple of (callable, data_keyword)..pipe will route the DataFrame to the argument specified in the tuple.

For example, we can fit a regression using statsmodels. Their API expects a formula first and a DataFrame as the second argument, data. We pass in the function, keyword pair (sm.ols, 'data') to pipe:

In [142]: import statsmodels.formula.api as sm

In [143]: bb = pd.read_csv('data/baseball.csv', index_col='id')

In [144]: (bb.query('h > 0') .....: .assign(ln_h=lambda df: np.log(df.h)) .....: .pipe((sm.ols, 'data'), 'hr ~ ln_h + year + g + C(lg)') .....: .fit() .....: .summary() .....: ) .....: Out[144]: <class 'statsmodels.iolib.summary.Summary'> """ OLS Regression Results

Dep. Variable: hr R-squared: 0.685 Model: OLS Adj. R-squared: 0.665 Method: Least Squares F-statistic: 34.28 Date: Tue, 12 Mar 2019 Prob (F-statistic): 3.48e-15 Time: 22:38:35 Log-Likelihood: -205.92 No. Observations: 68 AIC: 421.8 Df Residuals: 63 BIC: 432.9 Df Model: 4
Covariance Type: nonrobust 

              coef    std err          t      P>|t|      [0.025      0.975]

Intercept -8484.7720 4664.146 -1.819 0.074 -1.78e+04 835.780 C(lg)[T.NL] -2.2736 1.325 -1.716 0.091 -4.922 0.375 ln_h -1.3542 0.875 -1.547 0.127 -3.103 0.395 year 4.2277 2.324 1.819 0.074 -0.417 8.872 g 0.1841 0.029 6.258 0.000 0.125 0.243

Omnibus: 10.875 Durbin-Watson: 1.999 Prob(Omnibus): 0.004 Jarque-Bera (JB): 17.298 Skew: 0.537 Prob(JB): 0.000175 Kurtosis: 5.225 Cond. No. 1.49e+07

Warnings: [1] Standard Errors assume that the covariance matrix of the errors is correctly specified. [2] The condition number is large, 1.49e+07. This might indicate that there are strong multicollinearity or other numerical problems. """

The pipe method is inspired by unix pipes and more recently dplyr and magrittr, which have introduced the popular (%>%) (read pipe) operator for R. The implementation of pipe here is quite clean and feels right at home in python. We encourage you to view the source code of pipe().

Row or Column-wise Function Application

Arbitrary functions can be applied along the axes of a DataFrame using the apply() method, which, like the descriptive statistics methods, takes an optional axis argument:

In [145]: df.apply(np.mean) Out[145]: one 0.613869 two 0.470270 three -0.427633 dtype: float64

In [146]: df.apply(np.mean, axis=1) Out[146]: a -0.121116 b 0.361488 c 0.571564 d -0.058569 dtype: float64

In [147]: df.apply(lambda x: x.max() - x.min()) Out[147]: one 1.757280 two 3.196734 three 2.065853 dtype: float64

In [148]: df.apply(np.cumsum) Out[148]: one two three a 1.400810 -1.643041 NaN b 1.044340 -0.597130 0.395023 c 1.841608 0.327385 0.387933 d NaN 1.881078 -1.282898

In [149]: df.apply(np.exp) Out[149]: one two three a 4.058485 0.193391 NaN b 0.700143 2.845991 1.484418 c 2.219469 2.520646 0.992935 d NaN 4.728902 0.188091

The apply() method will also dispatch on a string method name.

In [150]: df.apply('mean') Out[150]: one 0.613869 two 0.470270 three -0.427633 dtype: float64

In [151]: df.apply('mean', axis=1) Out[151]: a -0.121116 b 0.361488 c 0.571564 d -0.058569 dtype: float64

The return type of the function passed to apply() affects the type of the final output from DataFrame.apply for the default behaviour:

This default behaviour can be overridden using the result_type, which accepts three options: reduce, broadcast, and expand. These will determine how list-likes return values expand (or not) to a DataFrame.

apply() combined with some cleverness can be used to answer many questions about a data set. For example, suppose we wanted to extract the date where the maximum value for each column occurred:

In [152]: tsdf = pd.DataFrame(np.random.randn(1000, 3), columns=['A', 'B', 'C'], .....: index=pd.date_range('1/1/2000', periods=1000)) .....:

In [153]: tsdf.apply(lambda x: x.idxmax()) Out[153]: A 2000-06-10 B 2001-07-04 C 2002-08-09 dtype: datetime64[ns]

You may also pass additional arguments and keyword arguments to the apply()method. For instance, consider the following function you would like to apply:

def subtract_and_divide(x, sub, divide=1): return (x - sub) / divide

You may then apply this function as follows:

df.apply(subtract_and_divide, args=(5,), divide=3)

Another useful feature is the ability to pass Series methods to carry out some Series operation on each column or row:

In [154]: tsdf Out[154]: A B C 2000-01-01 -0.652077 -0.239118 0.841272 2000-01-02 0.130224 0.347505 -0.385666 2000-01-03 -1.700237 -0.925899 0.199564 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.339319 -0.978307 0.689492 2000-01-09 0.601495 -0.630417 -1.040079 2000-01-10 1.511723 -0.427952 -0.400154

In [155]: tsdf.apply(pd.Series.interpolate) Out[155]: A B C 2000-01-01 -0.652077 -0.239118 0.841272 2000-01-02 0.130224 0.347505 -0.385666 2000-01-03 -1.700237 -0.925899 0.199564 2000-01-04 -1.292326 -0.936380 0.297550 2000-01-05 -0.884415 -0.946862 0.395535 2000-01-06 -0.476503 -0.957344 0.493521 2000-01-07 -0.068592 -0.967825 0.591507 2000-01-08 0.339319 -0.978307 0.689492 2000-01-09 0.601495 -0.630417 -1.040079 2000-01-10 1.511723 -0.427952 -0.400154

Finally, apply() takes an argument raw which is False by default, which converts each row or column into a Series before applying the function. When set to True, the passed function will instead receive an ndarray object, which has positive performance implications if you do not need the indexing functionality.

Aggregation API

New in version 0.20.0.

The aggregation API allows one to express possibly multiple aggregation operations in a single concise way. This API is similar across pandas objects, see groupby API, thewindow functions API, and the resample API. The entry point for aggregation is DataFrame.aggregate(), or the aliasDataFrame.agg().

We will use a similar starting frame from above:

In [156]: tsdf = pd.DataFrame(np.random.randn(10, 3), columns=['A', 'B', 'C'], .....: index=pd.date_range('1/1/2000', periods=10)) .....:

In [157]: tsdf.iloc[3:7] = np.nan

In [158]: tsdf Out[158]: A B C 2000-01-01 0.396575 -0.364907 0.051290 2000-01-02 -0.310517 -0.369093 -0.353151 2000-01-03 -0.522441 1.659115 -0.272364 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 -0.057890 1.148901 0.011528 2000-01-09 -0.155578 0.742150 0.107324 2000-01-10 0.531797 0.080254 0.833297

Using a single function is equivalent to apply(). You can also pass named methods as strings. These will return a Series of the aggregated output:

In [159]: tsdf.agg(np.sum) Out[159]: A -0.118055 B 2.896420 C 0.377923 dtype: float64

In [160]: tsdf.agg('sum') Out[160]: A -0.118055 B 2.896420 C 0.377923 dtype: float64

these are equivalent to a .sum() because we are aggregating

on a single function

In [161]: tsdf.sum() Out[161]: A -0.118055 B 2.896420 C 0.377923 dtype: float64

Single aggregations on a Series this will return a scalar value:

In [162]: tsdf.A.agg('sum') Out[162]: -0.11805495013260869

Aggregating with multiple functions

You can pass multiple aggregation arguments as a list. The results of each of the passed functions will be a row in the resulting DataFrame. These are naturally named from the aggregation function.

In [163]: tsdf.agg(['sum']) Out[163]: A B C sum -0.118055 2.89642 0.377923

Multiple functions yield multiple rows:

In [164]: tsdf.agg(['sum', 'mean']) Out[164]: A B C sum -0.118055 2.896420 0.377923 mean -0.019676 0.482737 0.062987

On a Series, multiple functions return a Series, indexed by the function names:

In [165]: tsdf.A.agg(['sum', 'mean']) Out[165]: sum -0.118055 mean -0.019676 Name: A, dtype: float64

Passing a lambda function will yield a <lambda> named row:

In [166]: tsdf.A.agg(['sum', lambda x: x.mean()]) Out[166]: sum -0.118055 -0.019676 Name: A, dtype: float64

Passing a named function will yield that name for the row:

In [167]: def mymean(x): .....: return x.mean() .....:

In [168]: tsdf.A.agg(['sum', mymean]) Out[168]: sum -0.118055 mymean -0.019676 Name: A, dtype: float64

Aggregating with a dict

Passing a dictionary of column names to a scalar or a list of scalars, to DataFrame.aggallows you to customize which functions are applied to which columns. Note that the results are not in any particular order, you can use an OrderedDict instead to guarantee ordering.

In [169]: tsdf.agg({'A': 'mean', 'B': 'sum'}) Out[169]: A -0.019676 B 2.896420 dtype: float64

Passing a list-like will generate a DataFrame output. You will get a matrix-like output of all of the aggregators. The output will consist of all unique functions. Those that are not noted for a particular column will be NaN:

In [170]: tsdf.agg({'A': ['mean', 'min'], 'B': 'sum'}) Out[170]: A B mean -0.019676 NaN min -0.522441 NaN sum NaN 2.89642

Mixed Dtypes

When presented with mixed dtypes that cannot aggregate, .agg will only take the valid aggregations. This is similar to how groupby .agg works.

In [171]: mdf = pd.DataFrame({'A': [1, 2, 3], .....: 'B': [1., 2., 3.], .....: 'C': ['foo', 'bar', 'baz'], .....: 'D': pd.date_range('20130101', periods=3)}) .....:

In [172]: mdf.dtypes Out[172]: A int64 B float64 C object D datetime64[ns] dtype: object

In [173]: mdf.agg(['min', 'sum']) Out[173]: A B C D min 1 1.0 bar 2013-01-01 sum 6 6.0 foobarbaz NaT

Custom describe

With .agg() is it possible to easily create a custom describe function, similar to the built in describe function.

In [174]: from functools import partial

In [175]: q_25 = partial(pd.Series.quantile, q=0.25)

In [176]: q_25.name = '25%'

In [177]: q_75 = partial(pd.Series.quantile, q=0.75)

In [178]: q_75.name = '75%'

In [179]: tsdf.agg(['count', 'mean', 'std', 'min', q_25, 'median', q_75, 'max']) Out[179]: A B C count 6.000000 6.000000 6.000000 mean -0.019676 0.482737 0.062987 std 0.408577 0.836785 0.420419 min -0.522441 -0.369093 -0.353151 25% -0.271782 -0.253617 -0.201391 median -0.106734 0.411202 0.031409 75% 0.282958 1.047213 0.093315 max 0.531797 1.659115 0.833297

Transform API

New in version 0.20.0.

The transform() method returns an object that is indexed the same (same size) as the original. This API allows you to provide multiple operations at the same time rather than one-by-one. Its API is quite similar to the .agg API.

We create a frame similar to the one used in the above sections.

In [180]: tsdf = pd.DataFrame(np.random.randn(10, 3), columns=['A', 'B', 'C'], .....: index=pd.date_range('1/1/2000', periods=10)) .....:

In [181]: tsdf.iloc[3:7] = np.nan

In [182]: tsdf Out[182]: A B C 2000-01-01 -1.219234 -1.652700 -0.698277 2000-01-02 1.858653 -0.738520 0.630364 2000-01-03 -0.112596 1.525897 1.364225 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 -0.527790 -1.715506 0.387274 2000-01-09 -0.569341 0.569386 0.134136 2000-01-10 -0.413993 -0.862280 0.662690

Transform the entire frame. .transform() allows input functions as: a NumPy function, a string function name or a user defined function.

In [183]: tsdf.transform(np.abs) Out[183]: A B C 2000-01-01 1.219234 1.652700 0.698277 2000-01-02 1.858653 0.738520 0.630364 2000-01-03 0.112596 1.525897 1.364225 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.527790 1.715506 0.387274 2000-01-09 0.569341 0.569386 0.134136 2000-01-10 0.413993 0.862280 0.662690

In [184]: tsdf.transform('abs') Out[184]: A B C 2000-01-01 1.219234 1.652700 0.698277 2000-01-02 1.858653 0.738520 0.630364 2000-01-03 0.112596 1.525897 1.364225 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.527790 1.715506 0.387274 2000-01-09 0.569341 0.569386 0.134136 2000-01-10 0.413993 0.862280 0.662690

In [185]: tsdf.transform(lambda x: x.abs()) Out[185]: A B C 2000-01-01 1.219234 1.652700 0.698277 2000-01-02 1.858653 0.738520 0.630364 2000-01-03 0.112596 1.525897 1.364225 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.527790 1.715506 0.387274 2000-01-09 0.569341 0.569386 0.134136 2000-01-10 0.413993 0.862280 0.662690

Here transform() received a single function; this is equivalent to a ufunc application.

In [186]: np.abs(tsdf) Out[186]: A B C 2000-01-01 1.219234 1.652700 0.698277 2000-01-02 1.858653 0.738520 0.630364 2000-01-03 0.112596 1.525897 1.364225 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.527790 1.715506 0.387274 2000-01-09 0.569341 0.569386 0.134136 2000-01-10 0.413993 0.862280 0.662690

Passing a single function to .transform() with a Series will yield a single Series in return.

In [187]: tsdf.A.transform(np.abs) Out[187]: 2000-01-01 1.219234 2000-01-02 1.858653 2000-01-03 0.112596 2000-01-04 NaN 2000-01-05 NaN 2000-01-06 NaN 2000-01-07 NaN 2000-01-08 0.527790 2000-01-09 0.569341 2000-01-10 0.413993 Freq: D, Name: A, dtype: float64

Transform with multiple functions

Passing multiple functions will yield a column MultiIndexed DataFrame. The first level will be the original frame column names; the second level will be the names of the transforming functions.

In [188]: tsdf.transform([np.abs, lambda x: x + 1]) Out[188]: A B C
absolute absolute absolute 2000-01-01 1.219234 -0.219234 1.652700 -0.652700 0.698277 0.301723 2000-01-02 1.858653 2.858653 0.738520 0.261480 0.630364 1.630364 2000-01-03 0.112596 0.887404 1.525897 2.525897 1.364225 2.364225 2000-01-04 NaN NaN NaN NaN NaN NaN 2000-01-05 NaN NaN NaN NaN NaN NaN 2000-01-06 NaN NaN NaN NaN NaN NaN 2000-01-07 NaN NaN NaN NaN NaN NaN 2000-01-08 0.527790 0.472210 1.715506 -0.715506 0.387274 1.387274 2000-01-09 0.569341 0.430659 0.569386 1.569386 0.134136 1.134136 2000-01-10 0.413993 0.586007 0.862280 0.137720 0.662690 1.662690

Passing multiple functions to a Series will yield a DataFrame. The resulting column names will be the transforming functions.

In [189]: tsdf.A.transform([np.abs, lambda x: x + 1]) Out[189]: absolute 2000-01-01 1.219234 -0.219234 2000-01-02 1.858653 2.858653 2000-01-03 0.112596 0.887404 2000-01-04 NaN NaN 2000-01-05 NaN NaN 2000-01-06 NaN NaN 2000-01-07 NaN NaN 2000-01-08 0.527790 0.472210 2000-01-09 0.569341 0.430659 2000-01-10 0.413993 0.586007

Transforming with a dict

Passing a dict of functions will allow selective transforming per column.

In [190]: tsdf.transform({'A': np.abs, 'B': lambda x: x + 1}) Out[190]: A B 2000-01-01 1.219234 -0.652700 2000-01-02 1.858653 0.261480 2000-01-03 0.112596 2.525897 2000-01-04 NaN NaN 2000-01-05 NaN NaN 2000-01-06 NaN NaN 2000-01-07 NaN NaN 2000-01-08 0.527790 -0.715506 2000-01-09 0.569341 1.569386 2000-01-10 0.413993 0.137720

Passing a dict of lists will generate a MultiIndexed DataFrame with these selective transforms.

In [191]: tsdf.transform({'A': np.abs, 'B': [lambda x: x + 1, 'sqrt']}) Out[191]: A B
absolute sqrt 2000-01-01 1.219234 -0.652700 NaN 2000-01-02 1.858653 0.261480 NaN 2000-01-03 0.112596 2.525897 1.235272 2000-01-04 NaN NaN NaN 2000-01-05 NaN NaN NaN 2000-01-06 NaN NaN NaN 2000-01-07 NaN NaN NaN 2000-01-08 0.527790 -0.715506 NaN 2000-01-09 0.569341 1.569386 0.754577 2000-01-10 0.413993 0.137720 NaN

Applying Elementwise Functions

Since not all functions can be vectorized (accept NumPy arrays and return another array or value), the methods applymap() on DataFrame and analogously map() on Series accept any Python function taking a single value and returning a single value. For example:

In [192]: df4 Out[192]: one two three a 1.400810 -1.643041 NaN b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 d NaN 1.553693 -1.670830

In [193]: def f(x): .....: return len(str(x)) .....:

In [194]: df4['one'].map(f) Out[194]: a 18 b 19 c 18 d 3 Name: one, dtype: int64

In [195]: df4.applymap(f) Out[195]: one two three a 18 19 3 b 19 18 19 c 18 18 21 d 3 18 19

Series.map() has an additional feature; it can be used to easily “link” or “map” values defined by a secondary series. This is closely related to merging/joining functionality:

In [196]: s = pd.Series(['six', 'seven', 'six', 'seven', 'six'], .....: index=['a', 'b', 'c', 'd', 'e']) .....:

In [197]: t = pd.Series({'six': 6., 'seven': 7.})

In [198]: s Out[198]: a six b seven c six d seven e six dtype: object

In [199]: s.map(t) Out[199]: a 6.0 b 7.0 c 6.0 d 7.0 e 6.0 dtype: float64

Reindexing and altering labels

reindex() is the fundamental data alignment method in pandas. It is used to implement nearly all other features relying on label-alignment functionality. To reindex means to conform the data to match a given set of labels along a particular axis. This accomplishes several things:

Here is a simple example:

In [200]: s = pd.Series(np.random.randn(5), index=['a', 'b', 'c', 'd', 'e'])

In [201]: s Out[201]: a -0.368437 b -0.036473 c 0.774830 d -0.310545 e 0.709717 dtype: float64

In [202]: s.reindex(['e', 'b', 'f', 'd']) Out[202]: e 0.709717 b -0.036473 f NaN d -0.310545 dtype: float64

Here, the f label was not contained in the Series and hence appears asNaN in the result.

With a DataFrame, you can simultaneously reindex the index and columns:

In [203]: df Out[203]: one two three a 1.400810 -1.643041 NaN b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 d NaN 1.553693 -1.670830

In [204]: df.reindex(index=['c', 'f', 'b'], columns=['three', 'two', 'one']) Out[204]: three two one c -0.007090 0.924515 0.797268 f NaN NaN NaN b 0.395023 1.045911 -0.356470

You may also use reindex with an axis keyword:

In [205]: df.reindex(['c', 'f', 'b'], axis='index') Out[205]: one two three c 0.797268 0.924515 -0.007090 f NaN NaN NaN b -0.356470 1.045911 0.395023

Note that the Index objects containing the actual axis labels can beshared between objects. So if we have a Series and a DataFrame, the following can be done:

In [206]: rs = s.reindex(df.index)

In [207]: rs Out[207]: a -0.368437 b -0.036473 c 0.774830 d -0.310545 dtype: float64

In [208]: rs.index is df.index Out[208]: True

This means that the reindexed Series’s index is the same Python object as the DataFrame’s index.

New in version 0.21.0.

DataFrame.reindex() also supports an “axis-style” calling convention, where you specify a single labels argument and the axis it applies to.

In [209]: df.reindex(['c', 'f', 'b'], axis='index') Out[209]: one two three c 0.797268 0.924515 -0.007090 f NaN NaN NaN b -0.356470 1.045911 0.395023

In [210]: df.reindex(['three', 'two', 'one'], axis='columns') Out[210]: three two one a NaN -1.643041 1.400810 b 0.395023 1.045911 -0.356470 c -0.007090 0.924515 0.797268 d -1.670830 1.553693 NaN

Note

When writing performance-sensitive code, there is a good reason to spend some time becoming a reindexing ninja: many operations are faster on pre-aligned data. Adding two unaligned DataFrames internally triggers a reindexing step. For exploratory analysis you will hardly notice the difference (because reindex has been heavily optimized), but when CPU cycles matter sprinkling a few explicit reindex calls here and there can have an impact.

Reindexing to align with another object

You may wish to take an object and reindex its axes to be labeled the same as another object. While the syntax for this is straightforward albeit verbose, it is a common enough operation that the reindex_like() method is available to make this simpler:

In [211]: df2 Out[211]: one two a 1.400810 -1.643041 b -0.356470 1.045911 c 0.797268 0.924515

In [212]: df3 Out[212]: one two a 0.786941 -1.752170 b -0.970339 0.936783 c 0.183399 0.815387

In [213]: df.reindex_like(df2) Out[213]: one two a 1.400810 -1.643041 b -0.356470 1.045911 c 0.797268 0.924515

Aligning objects with each other with align

The align() method is the fastest way to simultaneously align two objects. It supports a join argument (related to joining and merging):

It returns a tuple with both of the reindexed Series:

In [214]: s = pd.Series(np.random.randn(5), index=['a', 'b', 'c', 'd', 'e'])

In [215]: s1 = s[:4]

In [216]: s2 = s[1:]

In [217]: s1.align(s2) Out[217]: (a -0.610263 b -0.170883 c 0.367255 d 0.273860 e NaN dtype: float64, a NaN b -0.170883 c 0.367255 d 0.273860 e 0.314782 dtype: float64)

In [218]: s1.align(s2, join='inner') Out[218]: (b -0.170883 c 0.367255 d 0.273860 dtype: float64, b -0.170883 c 0.367255 d 0.273860 dtype: float64)

In [219]: s1.align(s2, join='left') Out[219]: (a -0.610263 b -0.170883 c 0.367255 d 0.273860 dtype: float64, a NaN b -0.170883 c 0.367255 d 0.273860 dtype: float64)

For DataFrames, the join method will be applied to both the index and the columns by default:

In [220]: df.align(df2, join='inner') Out[220]: ( one two a 1.400810 -1.643041 b -0.356470 1.045911 c 0.797268 0.924515, one two a 1.400810 -1.643041 b -0.356470 1.045911 c 0.797268 0.924515)

You can also pass an axis option to only align on the specified axis:

In [221]: df.align(df2, join='inner', axis=0) Out[221]: ( one two three a 1.400810 -1.643041 NaN b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090, one two a 1.400810 -1.643041 b -0.356470 1.045911 c 0.797268 0.924515)

If you pass a Series to DataFrame.align(), you can choose to align both objects either on the DataFrame’s index or columns using the axis argument:

In [222]: df.align(df2.iloc[0], axis=1) Out[222]: ( one three two a 1.400810 NaN -1.643041 b -0.356470 0.395023 1.045911 c 0.797268 -0.007090 0.924515 d NaN -1.670830 1.553693, one 1.400810 three NaN two -1.643041 Name: a, dtype: float64)

Filling while reindexing

reindex() takes an optional parameter method which is a filling method chosen from the following table:

Method Action
pad / ffill Fill values forward
bfill / backfill Fill values backward
nearest Fill from the nearest index value

We illustrate these fill methods on a simple Series:

In [223]: rng = pd.date_range('1/3/2000', periods=8)

In [224]: ts = pd.Series(np.random.randn(8), index=rng)

In [225]: ts2 = ts[[0, 3, 6]]

In [226]: ts Out[226]: 2000-01-03 -0.082578 2000-01-04 0.768554 2000-01-05 0.398842 2000-01-06 -0.357956 2000-01-07 0.156403 2000-01-08 -1.347564 2000-01-09 0.253506 2000-01-10 1.228964 Freq: D, dtype: float64

In [227]: ts2 Out[227]: 2000-01-03 -0.082578 2000-01-06 -0.357956 2000-01-09 0.253506 dtype: float64

In [228]: ts2.reindex(ts.index) Out[228]: 2000-01-03 -0.082578 2000-01-04 NaN 2000-01-05 NaN 2000-01-06 -0.357956 2000-01-07 NaN 2000-01-08 NaN 2000-01-09 0.253506 2000-01-10 NaN Freq: D, dtype: float64

In [229]: ts2.reindex(ts.index, method='ffill') Out[229]: 2000-01-03 -0.082578 2000-01-04 -0.082578 2000-01-05 -0.082578 2000-01-06 -0.357956 2000-01-07 -0.357956 2000-01-08 -0.357956 2000-01-09 0.253506 2000-01-10 0.253506 Freq: D, dtype: float64

In [230]: ts2.reindex(ts.index, method='bfill') Out[230]: 2000-01-03 -0.082578 2000-01-04 -0.357956 2000-01-05 -0.357956 2000-01-06 -0.357956 2000-01-07 0.253506 2000-01-08 0.253506 2000-01-09 0.253506 2000-01-10 NaN Freq: D, dtype: float64

In [231]: ts2.reindex(ts.index, method='nearest') Out[231]: 2000-01-03 -0.082578 2000-01-04 -0.082578 2000-01-05 -0.357956 2000-01-06 -0.357956 2000-01-07 -0.357956 2000-01-08 0.253506 2000-01-09 0.253506 2000-01-10 0.253506 Freq: D, dtype: float64

These methods require that the indexes are ordered increasing or decreasing.

Note that the same result could have been achieved usingfillna (except for method='nearest') orinterpolate:

In [232]: ts2.reindex(ts.index).fillna(method='ffill') Out[232]: 2000-01-03 -0.082578 2000-01-04 -0.082578 2000-01-05 -0.082578 2000-01-06 -0.357956 2000-01-07 -0.357956 2000-01-08 -0.357956 2000-01-09 0.253506 2000-01-10 0.253506 Freq: D, dtype: float64

reindex() will raise a ValueError if the index is not monotonically increasing or decreasing. fillna() and interpolate()will not perform any checks on the order of the index.

Limits on filling while reindexing

The limit and tolerance arguments provide additional control over filling while reindexing. Limit specifies the maximum count of consecutive matches:

In [233]: ts2.reindex(ts.index, method='ffill', limit=1) Out[233]: 2000-01-03 -0.082578 2000-01-04 -0.082578 2000-01-05 NaN 2000-01-06 -0.357956 2000-01-07 -0.357956 2000-01-08 NaN 2000-01-09 0.253506 2000-01-10 0.253506 Freq: D, dtype: float64

In contrast, tolerance specifies the maximum distance between the index and indexer values:

In [234]: ts2.reindex(ts.index, method='ffill', tolerance='1 day') Out[234]: 2000-01-03 -0.082578 2000-01-04 -0.082578 2000-01-05 NaN 2000-01-06 -0.357956 2000-01-07 -0.357956 2000-01-08 NaN 2000-01-09 0.253506 2000-01-10 0.253506 Freq: D, dtype: float64

Notice that when used on a DatetimeIndex, TimedeltaIndex orPeriodIndex, tolerance will coerced into a Timedelta if possible. This allows you to specify tolerance with appropriate strings.

Dropping labels from an axis

A method closely related to reindex is the drop() function. It removes a set of labels from an axis:

In [235]: df Out[235]: one two three a 1.400810 -1.643041 NaN b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 d NaN 1.553693 -1.670830

In [236]: df.drop(['a', 'd'], axis=0) Out[236]: one two three b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090

In [237]: df.drop(['one'], axis=1) Out[237]: two three a -1.643041 NaN b 1.045911 0.395023 c 0.924515 -0.007090 d 1.553693 -1.670830

Note that the following also works, but is a bit less obvious / clean:

In [238]: df.reindex(df.index.difference(['a', 'd'])) Out[238]: one two three b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090

Renaming / mapping labels

The rename() method allows you to relabel an axis based on some mapping (a dict or Series) or an arbitrary function.

In [239]: s Out[239]: a -0.610263 b -0.170883 c 0.367255 d 0.273860 e 0.314782 dtype: float64

In [240]: s.rename(str.upper) Out[240]: A -0.610263 B -0.170883 C 0.367255 D 0.273860 E 0.314782 dtype: float64

If you pass a function, it must return a value when called with any of the labels (and must produce a set of unique values). A dict or Series can also be used:

In [241]: df.rename(columns={'one': 'foo', 'two': 'bar'}, .....: index={'a': 'apple', 'b': 'banana', 'd': 'durian'}) .....: Out[241]: foo bar three apple 1.400810 -1.643041 NaN banana -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 durian NaN 1.553693 -1.670830

If the mapping doesn’t include a column/index label, it isn’t renamed. Note that extra labels in the mapping don’t throw an error.

New in version 0.21.0.

DataFrame.rename() also supports an “axis-style” calling convention, where you specify a single mapper and the axis to apply that mapping to.

In [242]: df.rename({'one': 'foo', 'two': 'bar'}, axis='columns') Out[242]: foo bar three a 1.400810 -1.643041 NaN b -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 d NaN 1.553693 -1.670830

In [243]: df.rename({'a': 'apple', 'b': 'banana', 'd': 'durian'}, axis='index') Out[243]: one two three apple 1.400810 -1.643041 NaN banana -0.356470 1.045911 0.395023 c 0.797268 0.924515 -0.007090 durian NaN 1.553693 -1.670830

The rename() method also provides an inplace named parameter that is by default False and copies the underlying data. Passinplace=True to rename the data in place.

New in version 0.18.0.

Finally, rename() also accepts a scalar or list-like for altering the Series.name attribute.

In [244]: s.rename("scalar-name") Out[244]: a -0.610263 b -0.170883 c 0.367255 d 0.273860 e 0.314782 Name: scalar-name, dtype: float64

New in version 0.24.0.

The methods rename_axis() and rename_axis()allow specific names of a MultiIndex to be changed (as opposed to the labels).

In [245]: df = pd.DataFrame({'x': [1, 2, 3, 4, 5, 6], .....: 'y': [10, 20, 30, 40, 50, 60]}, .....: index=pd.MultiIndex.from_product([['a', 'b', 'c'], [1, 2]], .....: names=['let', 'num'])) .....:

In [246]: df Out[246]: x y let num
a 1 1 10 2 2 20 b 1 3 30 2 4 40 c 1 5 50 2 6 60

In [247]: df.rename_axis(index={'let': 'abc'}) Out[247]: x y abc num
a 1 1 10 2 2 20 b 1 3 30 2 4 40 c 1 5 50 2 6 60

In [248]: df.rename_axis(index=str.upper) Out[248]: x y LET NUM
a 1 1 10 2 2 20 b 1 3 30 2 4 40 c 1 5 50 2 6 60

Iteration

The behavior of basic iteration over pandas objects depends on the type. When iterating over a Series, it is regarded as array-like, and basic iteration produces the values. Other data structures, like DataFrame and Panel, follow the dict-like convention of iterating over the “keys” of the objects.

In short, basic iteration (for i in object) produces:

Thus, for example, iterating over a DataFrame gives you the column names:

In [249]: df = pd.DataFrame({'col1': np.random.randn(3), .....: 'col2': np.random.randn(3)}, index=['a', 'b', 'c']) .....:

In [250]: for col in df: .....: print(col) .....: col1 col2

Pandas objects also have the dict-like iteritems() method to iterate over the (key, value) pairs.

To iterate over the rows of a DataFrame, you can use the following methods:

Warning

Iterating through pandas objects is generally slow. In many cases, iterating manually over the rows is not needed and can be avoided with one of the following approaches:

Warning

You should never modify something you are iterating over. This is not guaranteed to work in all cases. Depending on the data types, the iterator returns a copy and not a view, and writing to it will have no effect!

For example, in the following case setting the value has no effect:

In [251]: df = pd.DataFrame({'a': [1, 2, 3], 'b': ['a', 'b', 'c']})

In [252]: for index, row in df.iterrows(): .....: row['a'] = 10 .....:

In [253]: df Out[253]: a b 0 1 a 1 2 b 2 3 c

iteritems

Consistent with the dict-like interface, iteritems() iterates through key-value pairs:

For example:

In [254]: for item, frame in wp.iteritems(): .....: print(item) .....: print(frame) .....: Item1 A B C D 2000-01-01 -1.157892 -1.344312 0.844885 1.075770 2000-01-02 -0.109050 1.643563 -1.469388 0.357021 2000-01-03 -0.674600 -1.776904 -0.968914 -1.294524 2000-01-04 0.413738 0.276662 -0.472035 -0.013960 2000-01-05 -0.362543 -0.006154 -0.923061 0.895717 Item2 A B C D 2000-01-01 0.805244 -1.206412 2.565646 1.431256 2000-01-02 1.340309 -1.170299 -0.226169 0.410835 2000-01-03 0.813850 0.132003 -0.827317 -0.076467 2000-01-04 -1.187678 1.130127 -1.436737 -1.413681 2000-01-05 1.607920 1.024180 0.569605 0.875906

iterrows

iterrows() allows you to iterate through the rows of a DataFrame as Series objects. It returns an iterator yielding each index value along with a Series containing the data in each row:

In [255]: for row_index, row in df.iterrows(): .....: print(row_index, row, sep='\n') .....: 0 a 1 b a Name: 0, dtype: object 1 a 2 b b Name: 1, dtype: object 2 a 3 b c Name: 2, dtype: object

Note

Because iterrows() returns a Series for each row, it does not preserve dtypes across the rows (dtypes are preserved across columns for DataFrames). For example,

In [256]: df_orig = pd.DataFrame([[1, 1.5]], columns=['int', 'float'])

In [257]: df_orig.dtypes Out[257]: int int64 float float64 dtype: object

In [258]: row = next(df_orig.iterrows())[1]

In [259]: row Out[259]: int 1.0 float 1.5 Name: 0, dtype: float64

All values in row, returned as a Series, are now upcasted to floats, also the original integer value in column x:

In [260]: row['int'].dtype Out[260]: dtype('float64')

In [261]: df_orig['int'].dtype Out[261]: dtype('int64')

To preserve dtypes while iterating over the rows, it is better to use itertuples() which returns namedtuples of the values and which is generally much faster than iterrows().

For instance, a contrived way to transpose the DataFrame would be:

In [262]: df2 = pd.DataFrame({'x': [1, 2, 3], 'y': [4, 5, 6]})

In [263]: print(df2) x y 0 1 4 1 2 5 2 3 6

In [264]: print(df2.T) 0 1 2 x 1 2 3 y 4 5 6

In [265]: df2_t = pd.DataFrame({idx: values for idx, values in df2.iterrows()})

In [266]: print(df2_t) 0 1 2 x 1 2 3 y 4 5 6

itertuples

The itertuples() method will return an iterator yielding a namedtuple for each row in the DataFrame. The first element of the tuple will be the row’s corresponding index value, while the remaining values are the row values.

For instance:

In [267]: for row in df.itertuples(): .....: print(row) .....: Pandas(Index=0, a=1, b='a') Pandas(Index=1, a=2, b='b') Pandas(Index=2, a=3, b='c')

This method does not convert the row to a Series object; it merely returns the values inside a namedtuple. Therefore,itertuples() preserves the data type of the values and is generally faster as iterrows().

Note

The column names will be renamed to positional names if they are invalid Python identifiers, repeated, or start with an underscore. With a large number of columns (>255), regular tuples are returned.

.dt accessor

Series has an accessor to succinctly return datetime like properties for the_values_ of the Series, if it is a datetime/period like Series. This will return a Series, indexed like the existing Series.

datetime

In [268]: s = pd.Series(pd.date_range('20130101 09:10:12', periods=4))

In [269]: s Out[269]: 0 2013-01-01 09:10:12 1 2013-01-02 09:10:12 2 2013-01-03 09:10:12 3 2013-01-04 09:10:12 dtype: datetime64[ns]

In [270]: s.dt.hour Out[270]: 0 9 1 9 2 9 3 9 dtype: int64

In [271]: s.dt.second Out[271]: 0 12 1 12 2 12 3 12 dtype: int64

In [272]: s.dt.day Out[272]: 0 1 1 2 2 3 3 4 dtype: int64

This enables nice expressions like this:

In [273]: s[s.dt.day == 2] Out[273]: 1 2013-01-02 09:10:12 dtype: datetime64[ns]

You can easily produces tz aware transformations:

In [274]: stz = s.dt.tz_localize('US/Eastern')

In [275]: stz Out[275]: 0 2013-01-01 09:10:12-05:00 1 2013-01-02 09:10:12-05:00 2 2013-01-03 09:10:12-05:00 3 2013-01-04 09:10:12-05:00 dtype: datetime64[ns, US/Eastern]

In [276]: stz.dt.tz Out[276]: <DstTzInfo 'US/Eastern' LMT-1 day, 19:04:00 STD>

You can also chain these types of operations:

In [277]: s.dt.tz_localize('UTC').dt.tz_convert('US/Eastern') Out[277]: 0 2013-01-01 04:10:12-05:00 1 2013-01-02 04:10:12-05:00 2 2013-01-03 04:10:12-05:00 3 2013-01-04 04:10:12-05:00 dtype: datetime64[ns, US/Eastern]

You can also format datetime values as strings with Series.dt.strftime() which supports the same format as the standard strftime().

DatetimeIndex

In [278]: s = pd.Series(pd.date_range('20130101', periods=4))

In [279]: s Out[279]: 0 2013-01-01 1 2013-01-02 2 2013-01-03 3 2013-01-04 dtype: datetime64[ns]

In [280]: s.dt.strftime('%Y/%m/%d') Out[280]: 0 2013/01/01 1 2013/01/02 2 2013/01/03 3 2013/01/04 dtype: object

PeriodIndex

In [281]: s = pd.Series(pd.period_range('20130101', periods=4))

In [282]: s Out[282]: 0 2013-01-01 1 2013-01-02 2 2013-01-03 3 2013-01-04 dtype: period[D]

In [283]: s.dt.strftime('%Y/%m/%d') Out[283]: 0 2013/01/01 1 2013/01/02 2 2013/01/03 3 2013/01/04 dtype: object

The .dt accessor works for period and timedelta dtypes.

period

In [284]: s = pd.Series(pd.period_range('20130101', periods=4, freq='D'))

In [285]: s Out[285]: 0 2013-01-01 1 2013-01-02 2 2013-01-03 3 2013-01-04 dtype: period[D]

In [286]: s.dt.year Out[286]: 0 2013 1 2013 2 2013 3 2013 dtype: int64

In [287]: s.dt.day Out[287]: 0 1 1 2 2 3 3 4 dtype: int64

timedelta

In [288]: s = pd.Series(pd.timedelta_range('1 day 00:00:05', periods=4, freq='s'))

In [289]: s Out[289]: 0 1 days 00:00:05 1 1 days 00:00:06 2 1 days 00:00:07 3 1 days 00:00:08 dtype: timedelta64[ns]

In [290]: s.dt.days Out[290]: 0 1 1 1 2 1 3 1 dtype: int64

In [291]: s.dt.seconds Out[291]: 0 5 1 6 2 7 3 8 dtype: int64

In [292]: s.dt.components Out[292]: days hours minutes seconds milliseconds microseconds nanoseconds 0 1 0 0 5 0 0 0 1 1 0 0 6 0 0 0 2 1 0 0 7 0 0 0 3 1 0 0 8 0 0 0

Note

Series.dt will raise a TypeError if you access with a non-datetime-like values.

Vectorized string methods

Series is equipped with a set of string processing methods that make it easy to operate on each element of the array. Perhaps most importantly, these methods exclude missing/NA values automatically. These are accessed via the Series’sstr attribute and generally have names matching the equivalent (scalar) built-in string methods. For example:

In [293]: s = pd.Series(['A', 'B', 'C', 'Aaba', 'Baca', np.nan, 'CABA', 'dog', 'cat'])

In [294]: s.str.lower() Out[294]: 0 a 1 b 2 c 3 aaba 4 baca 5 NaN 6 caba 7 dog 8 cat dtype: object

Powerful pattern-matching methods are provided as well, but note that pattern-matching generally uses regular expressions by default (and in some cases always uses them).

Please see Vectorized String Methods for a complete description.

Sorting

Pandas supports three kinds of sorting: sorting by index labels, sorting by column values, and sorting by a combination of both.

By Index

The Series.sort_index() and DataFrame.sort_index() methods are used to sort a pandas object by its index levels.

In [295]: df = pd.DataFrame({ .....: 'one': pd.Series(np.random.randn(3), index=['a', 'b', 'c']), .....: 'two': pd.Series(np.random.randn(4), index=['a', 'b', 'c', 'd']), .....: 'three': pd.Series(np.random.randn(3), index=['b', 'c', 'd'])}) .....:

In [296]: unsorted_df = df.reindex(index=['a', 'd', 'c', 'b'], .....: columns=['three', 'two', 'one']) .....:

In [297]: unsorted_df Out[297]: three two one a NaN -0.867293 0.050162 d 1.215473 -0.051744 NaN c -0.421091 -0.712097 0.953102 b 1.205223 0.632624 -1.534113

DataFrame

In [298]: unsorted_df.sort_index() Out[298]: three two one a NaN -0.867293 0.050162 b 1.205223 0.632624 -1.534113 c -0.421091 -0.712097 0.953102 d 1.215473 -0.051744 NaN

In [299]: unsorted_df.sort_index(ascending=False) Out[299]: three two one d 1.215473 -0.051744 NaN c -0.421091 -0.712097 0.953102 b 1.205223 0.632624 -1.534113 a NaN -0.867293 0.050162

In [300]: unsorted_df.sort_index(axis=1) Out[300]: one three two a 0.050162 NaN -0.867293 d NaN 1.215473 -0.051744 c 0.953102 -0.421091 -0.712097 b -1.534113 1.205223 0.632624

Series

In [301]: unsorted_df['three'].sort_index() Out[301]: a NaN b 1.205223 c -0.421091 d 1.215473 Name: three, dtype: float64

By Values

The Series.sort_values() method is used to sort a Series by its values. TheDataFrame.sort_values() method is used to sort a DataFrame by its column or row values. The optional by parameter to DataFrame.sort_values() may used to specify one or more columns to use to determine the sorted order.

In [302]: df1 = pd.DataFrame({'one': [2, 1, 1, 1], .....: 'two': [1, 3, 2, 4], .....: 'three': [5, 4, 3, 2]}) .....:

In [303]: df1.sort_values(by='two') Out[303]: one two three 0 2 1 5 2 1 2 3 1 1 3 4 3 1 4 2

The by parameter can take a list of column names, e.g.:

In [304]: df1[['one', 'two', 'three']].sort_values(by=['one', 'two']) Out[304]: one two three 2 1 2 3 1 1 3 4 3 1 4 2 0 2 1 5

These methods have special treatment of NA values via the na_positionargument:

In [305]: s[2] = np.nan

In [306]: s.sort_values() Out[306]: 0 A 3 Aaba 1 B 4 Baca 6 CABA 8 cat 7 dog 2 NaN 5 NaN dtype: object

In [307]: s.sort_values(na_position='first') Out[307]: 2 NaN 5 NaN 0 A 3 Aaba 1 B 4 Baca 6 CABA 8 cat 7 dog dtype: object

By Indexes and Values

New in version 0.23.0.

Strings passed as the by parameter to DataFrame.sort_values() may refer to either columns or index level names.

Build MultiIndex

In [308]: idx = pd.MultiIndex.from_tuples([('a', 1), ('a', 2), ('a', 2), .....: ('b', 2), ('b', 1), ('b', 1)]) .....:

In [309]: idx.names = ['first', 'second']

Build DataFrame

In [310]: df_multi = pd.DataFrame({'A': np.arange(6, 0, -1)}, .....: index=idx) .....:

In [311]: df_multi Out[311]: A first second
a 1 6 2 5 2 4 b 2 3 1 2 1 1

Sort by ‘second’ (index) and ‘A’ (column)

In [312]: df_multi.sort_values(by=['second', 'A']) Out[312]: A first second
b 1 1 1 2 a 1 6 b 2 3 a 2 4 2 5

Note

If a string matches both a column name and an index level name then a warning is issued and the column takes precedence. This will result in an ambiguity error in a future version.

searchsorted

Series has the searchsorted() method, which works similarly tonumpy.ndarray.searchsorted().

In [313]: ser = pd.Series([1, 2, 3])

In [314]: ser.searchsorted([0, 3]) Out[314]: array([0, 2])

In [315]: ser.searchsorted([0, 4]) Out[315]: array([0, 3])

In [316]: ser.searchsorted([1, 3], side='right') Out[316]: array([1, 3])

In [317]: ser.searchsorted([1, 3], side='left') Out[317]: array([0, 2])

In [318]: ser = pd.Series([3, 1, 2])

In [319]: ser.searchsorted([0, 3], sorter=np.argsort(ser)) Out[319]: array([0, 2])

smallest / largest values

Series has the nsmallest() and nlargest() methods which return the smallest or largest \(n\) values. For a large Series this can be much faster than sorting the entire Series and calling head(n) on the result.

In [320]: s = pd.Series(np.random.permutation(10))

In [321]: s Out[321]: 0 5 1 3 2 2 3 0 4 7 5 6 6 9 7 1 8 4 9 8 dtype: int64

In [322]: s.sort_values() Out[322]: 3 0 7 1 2 2 1 3 8 4 0 5 5 6 4 7 9 8 6 9 dtype: int64

In [323]: s.nsmallest(3) Out[323]: 3 0 7 1 2 2 dtype: int64

In [324]: s.nlargest(3) Out[324]: 6 9 9 8 4 7 dtype: int64

DataFrame also has the nlargest and nsmallest methods.

In [325]: df = pd.DataFrame({'a': [-2, -1, 1, 10, 8, 11, -1], .....: 'b': list('abdceff'), .....: 'c': [1.0, 2.0, 4.0, 3.2, np.nan, 3.0, 4.0]}) .....:

In [326]: df.nlargest(3, 'a') Out[326]: a b c 5 11 f 3.0 3 10 c 3.2 4 8 e NaN

In [327]: df.nlargest(5, ['a', 'c']) Out[327]: a b c 5 11 f 3.0 3 10 c 3.2 4 8 e NaN 2 1 d 4.0 6 -1 f 4.0

In [328]: df.nsmallest(3, 'a') Out[328]: a b c 0 -2 a 1.0 1 -1 b 2.0 6 -1 f 4.0

In [329]: df.nsmallest(5, ['a', 'c']) Out[329]: a b c 0 -2 a 1.0 1 -1 b 2.0 6 -1 f 4.0 2 1 d 4.0 4 8 e NaN

Sorting by a MultiIndex column

You must be explicit about sorting when the column is a MultiIndex, and fully specify all levels to by.

In [330]: df1.columns = pd.MultiIndex.from_tuples([('a', 'one'), .....: ('a', 'two'), .....: ('b', 'three')]) .....:

In [331]: df1.sort_values(by=('a', 'two')) Out[331]: a b one two three 0 2 1 5 2 1 2 3 1 1 3 4 3 1 4 2

Copying

The copy() method on pandas objects copies the underlying data (though not the axis indexes, since they are immutable) and returns a new object. Note thatit is seldom necessary to copy objects. For example, there are only a handful of ways to alter a DataFrame in-place:

To be clear, no pandas method has the side effect of modifying your data; almost every method returns a new object, leaving the original object untouched. If the data is modified, it is because you did so explicitly.

dtypes

For the most part, pandas uses NumPy arrays and dtypes for Series or individual columns of a DataFrame. NumPy provides support for float,int, bool, timedelta64[ns] and datetime64[ns] (note that NumPy does not support timezone-aware datetimes).

Pandas and third-party libraries extend NumPy’s type system in a few places. This section describes the extensions pandas has made internally. See Extension Types for how to write your own extension that works with pandas. See Extension Data Types for a list of third-party libraries that have implemented an extension.

The following table lists all of pandas extension types. See the respective documentation sections for more on each type.

Kind of Data Data Type Scalar Array Documentation
tz-aware datetime DatetimeTZDtype Timestamp arrays.DatetimeArray Time Zone Handling
Categorical CategoricalDtype (none) Categorical Categorical Data
period (time spans) PeriodDtype Period arrays.PeriodArray Time Span Representation
sparse SparseDtype (none) arrays.SparseArray Sparse data structures
intervals IntervalDtype Interval arrays.IntervalArray IntervalIndex
nullable integer Int64Dtype, … (none) arrays.IntegerArray Nullable Integer Data Type

Pandas uses the object dtype for storing strings.

Finally, arbitrary objects may be stored using the object dtype, but should be avoided to the extent possible (for performance and interoperability with other libraries and methods. See object conversion).

A convenient dtypes attribute for DataFrame returns a Series with the data type of each column.

In [332]: dft = pd.DataFrame({'A': np.random.rand(3), .....: 'B': 1, .....: 'C': 'foo', .....: 'D': pd.Timestamp('20010102'), .....: 'E': pd.Series([1.0] * 3).astype('float32'), .....: 'F': False, .....: 'G': pd.Series([1] * 3, dtype='int8')}) .....:

In [333]: dft Out[333]: A B C D E F G 0 0.278831 1 foo 2001-01-02 1.0 False 1 1 0.242124 1 foo 2001-01-02 1.0 False 1 2 0.078031 1 foo 2001-01-02 1.0 False 1

In [334]: dft.dtypes Out[334]: A float64 B int64 C object D datetime64[ns] E float32 F bool G int8 dtype: object

On a Series object, use the dtype attribute.

In [335]: dft['A'].dtype Out[335]: dtype('float64')

If a pandas object contains data with multiple dtypes in a single column, the dtype of the column will be chosen to accommodate all of the data types (object is the most general).

these ints are coerced to floats

In [336]: pd.Series([1, 2, 3, 4, 5, 6.]) Out[336]: 0 1.0 1 2.0 2 3.0 3 4.0 4 5.0 5 6.0 dtype: float64

string data forces an object dtype

In [337]: pd.Series([1, 2, 3, 6., 'foo']) Out[337]: 0 1 1 2 2 3 3 6 4 foo dtype: object

The number of columns of each type in a DataFrame can be found by callingget_dtype_counts().

In [338]: dft.get_dtype_counts() Out[338]: float64 1 float32 1 int64 1 int8 1 datetime64[ns] 1 bool 1 object 1 dtype: int64

Numeric dtypes will propagate and can coexist in DataFrames. If a dtype is passed (either directly via the dtype keyword, a passed ndarray, or a passed Series, then it will be preserved in DataFrame operations. Furthermore, different numeric dtypes will NOT be combined. The following example will give you a taste.

In [339]: df1 = pd.DataFrame(np.random.randn(8, 1), columns=['A'], dtype='float32')

In [340]: df1 Out[340]: A 0 -1.641339 1 -0.314062 2 -0.679206 3 1.178243 4 0.181790 5 -2.044248 6 1.151282 7 -1.641398

In [341]: df1.dtypes Out[341]: A float32 dtype: object

In [342]: df2 = pd.DataFrame({'A': pd.Series(np.random.randn(8), dtype='float16'), .....: 'B': pd.Series(np.random.randn(8)), .....: 'C': pd.Series(np.array(np.random.randn(8), .....: dtype='uint8'))}) .....:

In [343]: df2 Out[343]: A B C 0 0.130737 -1.143729 1 1 0.289551 2.787500 0 2 0.590820 -0.708143 254 3 -0.020142 -1.512388 0 4 -1.048828 -0.243145 1 5 -0.808105 -0.650992 0 6 1.373047 2.090108 0 7 -0.254395 0.433098 0

In [344]: df2.dtypes Out[344]: A float16 B float64 C uint8 dtype: object

defaults

By default integer types are int64 and float types are float64,regardless of platform (32-bit or 64-bit). The following will all result in int64 dtypes.

In [345]: pd.DataFrame([1, 2], columns=['a']).dtypes Out[345]: a int64 dtype: object

In [346]: pd.DataFrame({'a': [1, 2]}).dtypes Out[346]: a int64 dtype: object

In [347]: pd.DataFrame({'a': 1}, index=list(range(2))).dtypes Out[347]: a int64 dtype: object

Note that Numpy will choose platform-dependent types when creating arrays. The following WILL result in int32 on 32-bit platform.

In [348]: frame = pd.DataFrame(np.array([1, 2]))

upcasting

Types can potentially be upcasted when combined with other types, meaning they are promoted from the current type (e.g. int to float).

In [349]: df3 = df1.reindex_like(df2).fillna(value=0.0) + df2

In [350]: df3 Out[350]: A B C 0 -1.510602 -1.143729 1.0 1 -0.024511 2.787500 0.0 2 -0.088385 -0.708143 254.0 3 1.158101 -1.512388 0.0 4 -0.867039 -0.243145 1.0 5 -2.852354 -0.650992 0.0 6 2.524329 2.090108 0.0 7 -1.895793 0.433098 0.0

In [351]: df3.dtypes Out[351]: A float32 B float64 C float64 dtype: object

DataFrame.to_numpy() will return the lower-common-denominator of the dtypes, meaning the dtype that can accommodate ALL of the types in the resulting homogeneous dtyped NumPy array. This can force some upcasting.

In [352]: df3.to_numpy().dtype Out[352]: dtype('float64')

astype

You can use the astype() method to explicitly convert dtypes from one to another. These will by default return a copy, even if the dtype was unchanged (pass copy=False to change this behavior). In addition, they will raise an exception if the astype operation is invalid.

Upcasting is always according to the numpy rules. If two different dtypes are involved in an operation, then the more general one will be used as the result of the operation.

In [353]: df3 Out[353]: A B C 0 -1.510602 -1.143729 1.0 1 -0.024511 2.787500 0.0 2 -0.088385 -0.708143 254.0 3 1.158101 -1.512388 0.0 4 -0.867039 -0.243145 1.0 5 -2.852354 -0.650992 0.0 6 2.524329 2.090108 0.0 7 -1.895793 0.433098 0.0

In [354]: df3.dtypes Out[354]: A float32 B float64 C float64 dtype: object

conversion of dtypes

In [355]: df3.astype('float32').dtypes Out[355]: A float32 B float32 C float32 dtype: object

Convert a subset of columns to a specified type using astype().

In [356]: dft = pd.DataFrame({'a': [1, 2, 3], 'b': [4, 5, 6], 'c': [7, 8, 9]})

In [357]: dft[['a', 'b']] = dft[['a', 'b']].astype(np.uint8)

In [358]: dft Out[358]: a b c 0 1 4 7 1 2 5 8 2 3 6 9

In [359]: dft.dtypes Out[359]: a uint8 b uint8 c int64 dtype: object

New in version 0.19.0.

Convert certain columns to a specific dtype by passing a dict to astype().

In [360]: dft1 = pd.DataFrame({'a': [1, 0, 1], 'b': [4, 5, 6], 'c': [7, 8, 9]})

In [361]: dft1 = dft1.astype({'a': np.bool, 'c': np.float64})

In [362]: dft1 Out[362]: a b c 0 True 4 7.0 1 False 5 8.0 2 True 6 9.0

In [363]: dft1.dtypes Out[363]: a bool b int64 c float64 dtype: object

Note

When trying to convert a subset of columns to a specified type using astype() and loc(), upcasting occurs.

loc() tries to fit in what we are assigning to the current dtypes, while [] will overwrite them taking the dtype from the right hand side. Therefore the following piece of code produces the unintended result.

In [364]: dft = pd.DataFrame({'a': [1, 2, 3], 'b': [4, 5, 6], 'c': [7, 8, 9]})

In [365]: dft.loc[:, ['a', 'b']].astype(np.uint8).dtypes Out[365]: a uint8 b uint8 dtype: object

In [366]: dft.loc[:, ['a', 'b']] = dft.loc[:, ['a', 'b']].astype(np.uint8)

In [367]: dft.dtypes Out[367]: a int64 b int64 c int64 dtype: object

object conversion

pandas offers various functions to try to force conversion of types from the object dtype to other types. In cases where the data is already of the correct type, but stored in an object array, theDataFrame.infer_objects() and Series.infer_objects() methods can be used to soft convert to the correct type.

In [368]: import datetime

In [369]: df = pd.DataFrame([[1, 2], .....: ['a', 'b'], .....: [datetime.datetime(2016, 3, 2), .....: datetime.datetime(2016, 3, 2)]]) .....:

In [370]: df = df.T

In [371]: df Out[371]: 0 1 2 0 1 a 2016-03-02 00:00:00 1 2 b 2016-03-02 00:00:00

In [372]: df.dtypes Out[372]: 0 object 1 object 2 object dtype: object

Because the data was transposed the original inference stored all columns as object, whichinfer_objects will correct.

In [373]: df.infer_objects().dtypes Out[373]: 0 int64 1 object 2 datetime64[ns] dtype: object

The following functions are available for one dimensional object arrays or scalars to perform hard conversion of objects to a specified type:

To force a conversion, we can pass in an errors argument, which specifies how pandas should deal with elements that cannot be converted to desired dtype or object. By default, errors='raise', meaning that any errors encountered will be raised during the conversion process. However, if errors='coerce', these errors will be ignored and pandas will convert problematic elements to pd.NaT (for datetime and timedelta) or np.nan (for numeric). This might be useful if you are reading in data which is mostly of the desired dtype (e.g. numeric, datetime), but occasionally has non-conforming elements intermixed that you want to represent as missing:

In [381]: import datetime

In [382]: m = ['apple', datetime.datetime(2016, 3, 2)]

In [383]: pd.to_datetime(m, errors='coerce') Out[383]: DatetimeIndex(['NaT', '2016-03-02'], dtype='datetime64[ns]', freq=None)

In [384]: m = ['apple', 2, 3]

In [385]: pd.to_numeric(m, errors='coerce') Out[385]: array([ nan, 2., 3.])

In [386]: m = ['apple', pd.Timedelta('1day')]

In [387]: pd.to_timedelta(m, errors='coerce') Out[387]: TimedeltaIndex([NaT, '1 days'], dtype='timedelta64[ns]', freq=None)

The errors parameter has a third option of errors='ignore', which will simply return the passed in data if it encounters any errors with the conversion to a desired data type:

In [388]: import datetime

In [389]: m = ['apple', datetime.datetime(2016, 3, 2)]

In [390]: pd.to_datetime(m, errors='ignore') Out[390]: Index(['apple', 2016-03-02 00:00:00], dtype='object')

In [391]: m = ['apple', 2, 3]

In [392]: pd.to_numeric(m, errors='ignore') Out[392]: array(['apple', 2, 3], dtype=object)

In [393]: m = ['apple', pd.Timedelta('1day')]

In [394]: pd.to_timedelta(m, errors='ignore') Out[394]: array(['apple', Timedelta('1 days 00:00:00')], dtype=object)

In addition to object conversion, to_numeric() provides another argument downcast, which gives the option of downcasting the newly (or already) numeric data to a smaller dtype, which can conserve memory:

In [395]: m = ['1', 2, 3]

In [396]: pd.to_numeric(m, downcast='integer') # smallest signed int dtype Out[396]: array([1, 2, 3], dtype=int8)

In [397]: pd.to_numeric(m, downcast='signed') # same as 'integer' Out[397]: array([1, 2, 3], dtype=int8)

In [398]: pd.to_numeric(m, downcast='unsigned') # smallest unsigned int dtype Out[398]: array([1, 2, 3], dtype=uint8)

In [399]: pd.to_numeric(m, downcast='float') # smallest float dtype Out[399]: array([ 1., 2., 3.], dtype=float32)

As these methods apply only to one-dimensional arrays, lists or scalars; they cannot be used directly on multi-dimensional objects such as DataFrames. However, with apply(), we can “apply” the function over each column efficiently:

In [400]: import datetime

In [401]: df = pd.DataFrame([ .....: ['2016-07-09', datetime.datetime(2016, 3, 2)]] * 2, dtype='O') .....:

In [402]: df Out[402]: 0 1 0 2016-07-09 2016-03-02 00:00:00 1 2016-07-09 2016-03-02 00:00:00

In [403]: df.apply(pd.to_datetime) Out[403]: 0 1 0 2016-07-09 2016-03-02 1 2016-07-09 2016-03-02

In [404]: df = pd.DataFrame([['1.1', 2, 3]] * 2, dtype='O')

In [405]: df Out[405]: 0 1 2 0 1.1 2 3 1 1.1 2 3

In [406]: df.apply(pd.to_numeric) Out[406]: 0 1 2 0 1.1 2 3 1 1.1 2 3

In [407]: df = pd.DataFrame([['5us', pd.Timedelta('1day')]] * 2, dtype='O')

In [408]: df Out[408]: 0 1 0 5us 1 days 00:00:00 1 5us 1 days 00:00:00

In [409]: df.apply(pd.to_timedelta) Out[409]: 0 1 0 00:00:00.000005 1 days 1 00:00:00.000005 1 days

gotchas

Performing selection operations on integer type data can easily upcast the data to floating. The dtype of the input data will be preserved in cases where nans are not introduced. See also Support for integer NA.

In [410]: dfi = df3.astype('int32')

In [411]: dfi['E'] = 1

In [412]: dfi Out[412]: A B C E 0 -1 -1 1 1 1 0 2 0 1 2 0 0 254 1 3 1 -1 0 1 4 0 0 1 1 5 -2 0 0 1 6 2 2 0 1 7 -1 0 0 1

In [413]: dfi.dtypes Out[413]: A int32 B int32 C int32 E int64 dtype: object

In [414]: casted = dfi[dfi > 0]

In [415]: casted Out[415]: A B C E 0 NaN NaN 1.0 1 1 NaN 2.0 NaN 1 2 NaN NaN 254.0 1 3 1.0 NaN NaN 1 4 NaN NaN 1.0 1 5 NaN NaN NaN 1 6 2.0 2.0 NaN 1 7 NaN NaN NaN 1

In [416]: casted.dtypes Out[416]: A float64 B float64 C float64 E int64 dtype: object

While float dtypes are unchanged.

In [417]: dfa = df3.copy()

In [418]: dfa['A'] = dfa['A'].astype('float32')

In [419]: dfa.dtypes Out[419]: A float32 B float64 C float64 dtype: object

In [420]: casted = dfa[df2 > 0]

In [421]: casted Out[421]: A B C 0 -1.510602 NaN 1.0 1 -0.024511 2.787500 NaN 2 -0.088385 NaN 254.0 3 NaN NaN NaN 4 NaN NaN 1.0 5 NaN NaN NaN 6 2.524329 2.090108 NaN 7 NaN 0.433098 NaN

In [422]: casted.dtypes Out[422]: A float32 B float64 C float64 dtype: object

Selecting columns based on dtype

The select_dtypes() method implements subsetting of columns based on their dtype.

First, let’s create a DataFrame with a slew of different dtypes:

In [423]: df = pd.DataFrame({'string': list('abc'), .....: 'int64': list(range(1, 4)), .....: 'uint8': np.arange(3, 6).astype('u1'), .....: 'float64': np.arange(4.0, 7.0), .....: 'bool1': [True, False, True], .....: 'bool2': [False, True, False], .....: 'dates': pd.date_range('now', periods=3), .....: 'category': pd.Series(list("ABC")).astype('category')}) .....:

In [424]: df['tdeltas'] = df.dates.diff()

In [425]: df['uint64'] = np.arange(3, 6).astype('u8')

In [426]: df['other_dates'] = pd.date_range('20130101', periods=3)

In [427]: df['tz_aware_dates'] = pd.date_range('20130101', periods=3, tz='US/Eastern')

In [428]: df Out[428]: string int64 uint8 float64 bool1 bool2 dates category tdeltas uint64 other_dates tz_aware_dates 0 a 1 3 4.0 True False 2019-03-12 22:38:38.692567 A NaT 3 2013-01-01 2013-01-01 00:00:00-05:00 1 b 2 4 5.0 False True 2019-03-13 22:38:38.692567 B 1 days 4 2013-01-02 2013-01-02 00:00:00-05:00 2 c 3 5 6.0 True False 2019-03-14 22:38:38.692567 C 1 days 5 2013-01-03 2013-01-03 00:00:00-05:00

And the dtypes:

In [429]: df.dtypes Out[429]: string object int64 int64 uint8 uint8 float64 float64 bool1 bool bool2 bool dates datetime64[ns] category category tdeltas timedelta64[ns] uint64 uint64 other_dates datetime64[ns] tz_aware_dates datetime64[ns, US/Eastern] dtype: object

select_dtypes() has two parameters include and exclude that allow you to say “give me the columns with these dtypes” (include) and/or “give the columns without these dtypes” (exclude).

For example, to select bool columns:

In [430]: df.select_dtypes(include=[bool]) Out[430]: bool1 bool2 0 True False 1 False True 2 True False

You can also pass the name of a dtype in the NumPy dtype hierarchy:

In [431]: df.select_dtypes(include=['bool']) Out[431]: bool1 bool2 0 True False 1 False True 2 True False

select_dtypes() also works with generic dtypes as well.

For example, to select all numeric and boolean columns while excluding unsigned integers:

In [432]: df.select_dtypes(include=['number', 'bool'], exclude=['unsignedinteger']) Out[432]: int64 float64 bool1 bool2 tdeltas 0 1 4.0 True False NaT 1 2 5.0 False True 1 days 2 3 6.0 True False 1 days

To select string columns you must use the object dtype:

In [433]: df.select_dtypes(include=['object']) Out[433]: string 0 a 1 b 2 c

To see all the child dtypes of a generic dtype like numpy.number you can define a function that returns a tree of child dtypes:

In [434]: def subdtypes(dtype): .....: subs = dtype.subclasses() .....: if not subs: .....: return dtype .....: return [dtype, [subdtypes(dt) for dt in subs]] .....:

All NumPy dtypes are subclasses of numpy.generic:

In [435]: subdtypes(np.generic) Out[435]: [numpy.generic, [[numpy.number, [[numpy.integer, [[numpy.signedinteger, [numpy.int8, numpy.int16, numpy.int32, numpy.int64, numpy.int64, numpy.timedelta64]], [numpy.unsignedinteger, [numpy.uint8, numpy.uint16, numpy.uint32, numpy.uint64, numpy.uint64]]]], [numpy.inexact, [[numpy.floating, [numpy.float16, numpy.float32, numpy.float64, numpy.float128]], [numpy.complexfloating, [numpy.complex64, numpy.complex128, numpy.complex256]]]]]], [numpy.flexible, [[numpy.character, [numpy.bytes_, numpy.str_]], [numpy.void, [numpy.record]]]], numpy.bool_, numpy.datetime64, numpy.object_]]

Note

Pandas also defines the types category, and datetime64[ns, tz], which are not integrated into the normal NumPy hierarchy and won’t show up with the above function.