Pandas学习之差分函数diff

在上一篇的文章中，我们学习了Pandas的shift函数，今天要来学习的是diff函数，shift函数与diff函数有着莫大的关联，先来看看diff函数的官方说明：

>>> import pandas
>>> help(pandas.DataFrame.diff)
Help on function diff in module pandas.core.frame:
 
diff(self, periods=1, axis=0)
    First discrete difference of element.
 
    Calculates the difference of a DataFrame element compared with another
    element in the DataFrame (default is the element in the same column
    of the previous row).
 
    Parameters
    ----------
    periods : int, default 1
        Periods to shift for calculating difference, accepts negative
        values.
    axis : {0 or 'index', 1 or 'columns'}, default 0
        Take difference over rows (0) or columns (1).
 
        .. versionadded:: 0.16.1.
 
    Returns
    -------
    diffed : DataFrame
 
    See Also
    --------
    Series.diff: First discrete difference for a Series.
    DataFrame.pct_change: Percent change over given number of periods.
    DataFrame.shift: Shift index by desired number of periods with an
        optional time freq.
 
    Examples
    --------
    Difference with previous row
 
    >>> df = pd.DataFrame({'a': [1, 2, 3, 4, 5, 6],
    ...                    'b': [1, 1, 2, 3, 5, 8],
    ...                    'c': [1, 4, 9, 16, 25, 36]})
    >>> df
       a  b   c
    0  1  1   1
    1  2  1   4
    2  3  2   9
    3  4  3  16
    4  5  5  25
    5  6  8  36
 
    >>> df.diff()
         a    b     c
    0  NaN  NaN   NaN
    1  1.0  0.0   3.0
    2  1.0  1.0   5.0
    3  1.0  1.0   7.0
    4  1.0  2.0   9.0
    5  1.0  3.0  11.0
 
    Difference with previous column
 
    >>> df.diff(axis=1)
        a    b     c
    0 NaN  0.0   0.0
    1 NaN -1.0   3.0
    2 NaN -1.0   7.0
    3 NaN -1.0  13.0
    4 NaN  0.0  20.0
    5 NaN  2.0  28.0
 
    Difference with 3rd previous row
 
    >>> df.diff(periods=3)
         a    b     c
    0  NaN  NaN   NaN
    1  NaN  NaN   NaN
    2  NaN  NaN   NaN
    3  3.0  2.0  15.0
    4  3.0  4.0  21.0
    5  3.0  6.0  27.0
 
    Difference with following row
 
    >>> df.diff(periods=-1)
         a    b     c
    0 -1.0  0.0  -3.0
    1 -1.0 -1.0  -5.0
    2 -1.0 -1.0  -7.0
    3 -1.0 -2.0  -9.0
    4 -1.0 -3.0 -11.0
    5  NaN  NaN   NaN

从官方的说明中已经很明确的可以知道其shift函数的关系为：df.diff() = df – df.shift()

diff相比shift少了一个freq参数，函数原型为：diff(self, periods=1, axis=0)

其参数含义为：

• periods：移动的幅度，int类型，默认值为1。
• axis：移动的方向，{0 or ‘index’, 1 or ‘columns’}，如果为0或者’index’，则上下移动，如果为1或者’columns’，则左右移动。