Frequently Asked Questions (FAQ)

DataFrame memory usage

The memory usage of a DataFrame (including the index) is shown when callingthe info(). A configuration option, display.memory_usage(see the list of options), specifies if theDataFrame’s memory usage will be displayed when invoking the df.info()method.

For example, the memory usage of the DataFrame below is shownwhen calling info():

In [1]: dtypes = ['int64', 'float64', 'datetime64[ns]', 'timedelta64[ns]',
   ...:           'complex128', 'object', 'bool']
   ...: 
 
In [2]: n = 5000
 
In [3]: data = {t: np.random.randint(100, size=n).astype(t) for t in dtypes}
 
In [4]: df = pd.DataFrame(data)
 
In [5]: df['categorical'] = df['object'].astype('category')
 
In [6]: df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5000 entries, 0 to 4999
Data columns (total 8 columns):
int64              5000 non-null int64
float64            5000 non-null float64
datetime64[ns]     5000 non-null datetime64[ns]
timedelta64[ns]    5000 non-null timedelta64[ns]
complex128         5000 non-null complex128
object             5000 non-null object
bool               5000 non-null bool
categorical        5000 non-null category
dtypes: bool(1), category(1), complex128(1), datetime64[ns](1), float64(1), int64(1), object(1), timedelta64[ns](1)
memory usage: 289.1+ KB

The + symbol indicates that the true memory usage could be higher, becausepandas does not count the memory used by values in columns withdtype=object.

Passing memory_usage='deep' will enable a more accurate memory usage report,accounting for the full usage of the contained objects. This is optionalas it can be expensive to do this deeper introspection.

In [7]: df.info(memory_usage='deep')
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5000 entries, 0 to 4999
Data columns (total 8 columns):
int64              5000 non-null int64
float64            5000 non-null float64
datetime64[ns]     5000 non-null datetime64[ns]
timedelta64[ns]    5000 non-null timedelta64[ns]
complex128         5000 non-null complex128
object             5000 non-null object
bool               5000 non-null bool
categorical        5000 non-null category
dtypes: bool(1), category(1), complex128(1), datetime64[ns](1), float64(1), int64(1), object(1), timedelta64[ns](1)
memory usage: 425.6 KB

By default the display option is set to True but can be explicitlyoverridden by passing the memory_usage argument when invoking df.info().

The memory usage of each column can be found by calling thememory_usage() method. This returns a Series with an indexrepresented by column names and memory usage of each column shown in bytes. Forthe DataFrame above, the memory usage of each column and the total memoryusage can be found with the memory_usage method:

In [8]: df.memory_usage()
Out[8]: 
Index                128
int64              40000
float64            40000
datetime64[ns]     40000
timedelta64[ns]    40000
complex128         80000
object             40000
bool                5000
categorical        10920
dtype: int64
 
# total memory usage of dataframe
In [9]: df.memory_usage().sum()
Out[9]: 296048

By default the memory usage of the DataFrame’s index is shown in thereturned Series, the memory usage of the index can be suppressed by passingthe index=False argument:

In [10]: df.memory_usage(index=False)
Out[10]: 
int64              40000
float64            40000
datetime64[ns]     40000
timedelta64[ns]    40000
complex128         80000
object             40000
bool                5000
categorical        10920
dtype: int64

The memory usage displayed by the info() method utilizes thememory_usage() method to determine the memory usage of aDataFrame while also formatting the output in human-readable units (base-2representation; i.e. 1KB = 1024 bytes).

See also Categorical Memory Usage.

Using if/truth statements with pandas

pandas follows the NumPy convention of raising an error when you try to convertsomething to a bool. This happens in an if-statement or when using theboolean operations: and, or, and not. It is not clear what the resultof the following code should be:

>>> if pd.Series([False, True, False]):
...     pass

Should it be True because it’s not zero-length, or False because thereare False values? It is unclear, so instead, pandas raises a ValueError:

>>> if pd.Series([False, True, False]):
...     print("I was true")
Traceback
    ...
ValueError: The truth value of an array is ambiguous. Use a.empty, a.any() or a.all().

You need to explicitly choose what you want to do with the DataFrame, e.g.use any(), all() or empty().Alternatively, you might want to compare if the pandas object is None:

>>> if pd.Series([False, True, False]) is not None:
...     print("I was not None")
I was not None

Below is how to check if any of the values are True:

>>> if pd.Series([False, True, False]).any():
...     print("I am any")
I am any

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

In [11]: pd.Series([True]).bool()
Out[11]: True
 
In [12]: pd.Series([False]).bool()
Out[12]: False
 
In [13]: pd.DataFrame([[True]]).bool()
Out[13]: True
 
In [14]: pd.DataFrame([[False]]).bool()
Out[14]: False

Bitwise boolean

Bitwise boolean operators like == and != return a boolean Series,which is almost always what you want anyways.

>>> s = pd.Series(range(5))
>>> s == 4
0    False
1    False
2    False
3    False
4     True
dtype: bool

See boolean comparisons for more examples.

Using the in operator

Using the Python in operator on a Series tests for membership in theindex, not membership among the values.

In [15]: s = pd.Series(range(5), index=list('abcde'))
 
In [16]: 2 in s
Out[16]: False
 
In [17]: 'b' in s
Out[17]: True

If this behavior is surprising, keep in mind that using in on a Pythondictionary tests keys, not values, and Series are dict-like.To test for membership in the values, use the method isin():

In [18]: s.isin([2])
Out[18]: 
a    False
b    False
c     True
d    False
e    False
dtype: bool
 
In [19]: s.isin([2]).any()
Out[19]: True

For DataFrames, likewise, in applies to the column axis,testing for membership in the list of column names.

NaN, Integer NA values and NA type promotions

Choice of NA representation

For lack of NA (missing) support from the ground up in NumPy and Python ingeneral, we were given the difficult choice between either:

• A masked array solution: an array of data and an array of boolean valuesindicating whether a value is there or is missing.
• Using a special sentinel value, bit pattern, or set of sentinel values todenote NA across the dtypes.

For many reasons we chose the latter. After years of production use it hasproven, at least in my opinion, to be the best decision given the state ofaffairs in NumPy and Python in general. The special value NaN(Not-A-Number) is used everywhere as the NA value, and there are APIfunctions isna and notna which can be used across the dtypes todetect NA values.

However, it comes with it a couple of trade-offs which I most certainly havenot ignored.

Support for integer NA

In the absence of high performance NA support being built into NumPy fromthe ground up, the primary casualty is the ability to represent NAs in integerarrays. For example:

In [20]: s = pd.Series([1, 2, 3, 4, 5], index=list('abcde'))
 
In [21]: s
Out[21]: 
a    1
b    2
c    3
d    4
e    5
dtype: int64
 
In [22]: s.dtype
Out[22]: dtype('int64')
 
In [23]: s2 = s.reindex(['a', 'b', 'c', 'f', 'u'])
 
In [24]: s2
Out[24]: 
a    1.0
b    2.0
c    3.0
f    NaN
u    NaN
dtype: float64
 
In [25]: s2.dtype
Out[25]: dtype('float64')

This trade-off is made largely for memory and performance reasons, and also sothat the resulting Series continues to be “numeric”.

If you need to represent integers with possibly missing values, use one ofthe nullable-integer extension dtypes provided by pandas

• Int8Dtype
• Int16Dtype
• Int32Dtype
• Int64Dtype

In [26]: s_int = pd.Series([1, 2, 3, 4, 5], index=list('abcde'),
   ....:                   dtype=pd.Int64Dtype())
   ....: 
 
In [27]: s_int
Out[27]: 
a    1
b    2
c    3
d    4
e    5
dtype: Int64
 
In [28]: s_int.dtype
Out[28]: Int64Dtype()
 
In [29]: s2_int = s_int.reindex(['a', 'b', 'c', 'f', 'u'])
 
In [30]: s2_int
Out[30]: 
a      1
b      2
c      3
f    NaN
u    NaN
dtype: Int64
 
In [31]: s2_int.dtype
Out[31]: Int64Dtype()

See Nullable integer data type for more.

NA type promotions

When introducing NAs into an existing Series or DataFrame viareindex() or some other means, boolean and integer types will bepromoted to a different dtype in order to store the NAs. The promotions aresummarized in this table:

While this may seem like a heavy trade-off, I have found very few cases wherethis is an issue in practice i.e. storing values greater than 2**53. Someexplanation for the motivation is in the next section.

Why not make NumPy like R?

Many people have suggested that NumPy should simply emulate the NA supportpresent in the more domain-specific statistical programming language R. Part of the reason is the NumPy type hierarchy:

The R language, by contrast, only has a handful of built-in data types:integer, numeric (floating-point), character, andboolean. NA types are implemented by reserving special bit patterns foreach type to be used as the missing value. While doing this with the full NumPytype hierarchy would be possible, it would be a more substantial trade-off(especially for the 8- and 16-bit data types) and implementation undertaking.

An alternate approach is that of using masked arrays. A masked array is anarray of data with an associated boolean mask denoting whether each valueshould be considered NA or not. I am personally not in love with thisapproach as I feel that overall it places a fairly heavy burden on the user andthe library implementer. Additionally, it exacts a fairly high performance costwhen working with numerical data compared with the simple approach of usingNaN. Thus, I have chosen the Pythonic “practicality beats purity” approachand traded integer NA capability for a much simpler approach of using aspecial value in float and object arrays to denote NA, and promotinginteger arrays to floating when NAs must be introduced.

Differences with NumPy

For Series and DataFrame objects, var() normalizes byN-1 to produce unbiased estimates of the sample variance, while NumPy’svar normalizes by N, which measures the variance of the sample. Note thatcov() normalizes by N-1 in both pandas and NumPy.

Thread-safety

As of pandas 0.11, pandas is not 100% thread safe. The known issues relate tothe copy() method. If you are doing a lot of copying ofDataFrame objects shared among threads, we recommend holding locks insidethe threads where the data copying occurs.

See this linkfor more information.

Byte-Ordering issues

Occasionally you may have to deal with data that were created on a machine witha different byte order than the one on which you are running Python. A commonsymptom of this issue is an error like::

Traceback
    ...
ValueError: Big-endian buffer not supported on little-endian compiler

To dealwith this issue you should convert the underlying NumPy array to the nativesystem byte order before passing it to Series or DataFrameconstructors using something similar to the following:

In [32]: x = np.array(list(range(10)), '>i4')  # big endian
 
In [33]: newx = x.byteswap().newbyteorder()  # force native byteorder
 
In [34]: s = pd.Series(newx)

See the NumPy documentation on byte order for moredetails.