numpy基礎知識(一)

建立ndarray

In [1]: data = [[1,2,3,4],[5,6,7,8]]

In [2]: arry = np.array(data)

In [3]: arry
Out[3]: 
array([[1, 2, 3, 4],
       [5, 6, 7, 8]])

查看ndarray的維度

In [4]: arry.ndim
Out[4]: 2

輸出ndarray的行數和列數

In [5]: arry.shape
Out[5]: (2, 4)

zeros和ones能夠建立指定長度或形狀的全是零和一的數組

In [15]: np.zeros((3,4))
Out[15]: 
array([[0., 0., 0., 0.],
       [0., 0., 0., 0.],
       [0., 0., 0., 0.]])

empty會產生一些未初始化的垃圾值

In [17]: np.empty((2,3,2))
Out[17]: 
array([[[0., 0.],
        [0., 0.],
        [0., 0.]],

       [[0., 0.],
        [0., 0.],
        [0., 0.]]])

數組和標量之間的運算

In [1]: arry = np.array([[1,2,3,4],[5,6,7,8]])

In [2]: arry * arry
Out[2]: 
array([[ 1,  4,  9, 16],
       [25, 36, 49, 64]])

In [3]: arry - arry
Out[3]: 
array([[0, 0, 0, 0],
       [0, 0, 0, 0]])
In [4]: 1 / arry
Out[4]: 
array([[1, 0, 0, 0],
       [0, 0, 0, 0]])

In [5]: arry ** 0.5
Out[5]: 
array([[1.        , 1.41421356, 1.73205081, 2.        ],
       [2.23606798, 2.44948974, 2.64575131, 2.82842712]])

基本索引和切片

In [6]: arr = np.arange(15)

In [7]: arr
Out[7]: array([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14])

In [8]: arr[5]
Out[8]: 5

In [9]: arr[5:8]
Out[9]: array([5, 6, 7])
In [10]: arr[5:8] = 12

In [11]: arr
Out[11]: array([ 0,  1,  2,  3,  4, 12, 12, 12,  8,  9, 10, 11, 12, 13, 14])
In [12]: arr_slice = arr[5:8]

In [13]: arr_slice[1] = 1234

In [14]: arr
Out[14]: 
array([   0,    1,    2,    3,    4,   12, 1234,   12,    8,    9,   10,
         11,   12,   13,   14])
In [19]: arr2d = np.array([[1,2,3],[4,5,6],[7,8,9]])
> 將一個標量賦值給一個切片時，該值將自動傳播到選中的區域。跟列表最重要的區別在於，數組切片是原始數組的視圖，這意味着數據不會被複制，視圖上的修改會直接反應到原始數組。主要緣由若是numpy堅持將數據複製來複制去會產生性能和內存問題
In [20]: arr2d
Out[20]: 
array([[1, 2, 3],
       [4, 5, 6],
       [7, 8, 9]])

In [21]: arr2d[:2]
Out[21]: 
array([[1, 2, 3],
       [4, 5, 6]])

In [22]: arr2d[:2,:1]
Out[22]: 
array([[1],
       [4]])
In [23]: arr2d[:2,1:]
Out[23]: 
array([[2, 3],
       [5, 6]])
In [24]: arr2d[:,:1]
Out[24]: 
array([[1],
       [4],
       [7]])
>能夠一次性傳入多個切片;只有冒號時，表示選取整個軸

布爾型索引

In [29]: names
Out[29]: array(['Bob', 'Joe', 'Will', 'Bob', 'Will', 'Joe', 'Joe'], dtype='|S4')
In [33]: data = np.random.randn(7,4)

In [34]: data
Out[34]: 
array([[ 1.16675978,  1.92107701,  0.26941017, -1.19809428],
       [ 0.07649339, -0.91997008, -0.21832289,  1.07434831],
       [-1.98754599, -0.60850528,  0.06302264, -0.63377773],
       [ 0.32234302,  0.32310855,  0.73595001,  0.24015727],
       [-1.09218583, -0.08907211, -1.51447014,  0.53083635],
       [-0.27048015, -0.09299773,  0.75937323, -0.71705501],
       [ 1.23737272, -0.08242024, -2.3591149 ,  0.64462961]])
In [37]: data[names=='Bob']
Out[37]: 
array([[ 1.16675978,  1.92107701,  0.26941017, -1.19809428],
       [ 0.32234302,  0.32310855,  0.73595001,  0.24015727]])
>布爾型數組的長度必須和被索引軸相同
In [38]: data[names=='Bob',2:]
Out[38]: 
array([[ 0.26941017, -1.19809428],
       [ 0.73595001,  0.24015727]])
In [40]: data[names != 'Bob']
Out[40]: 
array([[ 0.07649339, -0.91997008, -0.21832289,  1.07434831],
       [-1.98754599, -0.60850528,  0.06302264, -0.63377773],
       [-1.09218583, -0.08907211, -1.51447014,  0.53083635],
       [-0.27048015, -0.09299773,  0.75937323, -0.71705501],
       [ 1.23737272, -0.08242024, -2.3591149 ,  0.64462961]])
In [41]: mask = (names == 'Bob') | (names == 'Will')

In [42]: data[mask]
Out[42]: 
array([[ 1.16675978,  1.92107701,  0.26941017, -1.19809428],
       [-1.98754599, -0.60850528,  0.06302264, -0.63377773],
       [ 0.32234302,  0.32310855,  0.73595001,  0.24015727],
       [-1.09218583, -0.08907211, -1.51447014,  0.53083635]])
>支持與或非的布爾運算
In [43]: data[data<0] = 0

In [44]: data
Out[44]: 
array([[1.16675978, 1.92107701, 0.26941017, 0.        ],
       [0.07649339, 0.        , 0.        , 1.07434831],
       [0.        , 0.        , 0.06302264, 0.        ],
       [0.32234302, 0.32310855, 0.73595001, 0.24015727],
       [0.        , 0.        , 0.        , 0.53083635],
       [0.        , 0.        , 0.75937323, 0.        ],
       [1.23737272, 0.        , 0.        , 0.64462961]])

In [45]:
>把小於零的值製成零

花式索引

In [45]: arr = np.empty((8,4))
In [48]: for i in range(8):
    ...:     arr[i] = i
    ...:     

In [49]: arr
Out[49]: 
array([[0., 0., 0., 0.],
       [1., 1., 1., 1.],
       [2., 2., 2., 2.],
       [3., 3., 3., 3.],
       [4., 4., 4., 4.],
       [5., 5., 5., 5.],
       [6., 6., 6., 6.],
       [7., 7., 7., 7.]])

In [50]: arr[[4,3,0,6]]
Out[50]: 
array([[4., 4., 4., 4.],
       [3., 3., 3., 3.],
       [0., 0., 0., 0.],
       [6., 6., 6., 6.]])
In [51]: arr[[-3,-5,-7]]
Out[51]: 
array([[5., 5., 5., 5.],
       [3., 3., 3., 3.],
       [1., 1., 1., 1.]])
>負數索引是從末尾開始選取
In [52]: arr = np.arange(32).reshape((8,4))

In [53]: arr
Out[53]: 
array([[ 0,  1,  2,  3],
       [ 4,  5,  6,  7],
       [ 8,  9, 10, 11],
       [12, 13, 14, 15],
       [16, 17, 18, 19],
       [20, 21, 22, 23],
       [24, 25, 26, 27],
       [28, 29, 30, 31]])
In [54]: arr[[1,5,7,2],[0,3,1,2]]
Out[54]: array([ 4, 23, 29, 10])
>這種結果不是咱們所須要的
In [55]: arr[[1,5,7,2]][:,[0,3,1,2]]
Out[55]: 
array([[ 4,  7,  5,  6],
       [20, 23, 21, 22],
       [28, 31, 29, 30],
       [ 8, 11,  9, 10]]
In [57]: arr[np.ix_([1,5,7,2],[0,3,1,2])]
Out[57]: 
array([[ 4,  7,  5,  6],
       [20, 23, 21, 22],
       [28, 31, 29, 30],
       [ 8, 11,  9, 10]])
>np.ix_函數能夠把一位函數變成索引選擇器

數組轉置和軸對換

In [60]: arr
Out[60]: 
array([[ 0,  1,  2,  3,  4],
       [ 5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14]])

In [61]: arr.T
Out[61]: 
array([[ 0,  5, 10],
       [ 1,  6, 11],
       [ 2,  7, 12],
       [ 3,  8, 13],
       [ 4,  9, 14]])

相關概念

1. 矢量化:即便不用編寫循環便可對數據進行計算.
2.不一樣大小數組之間的運算叫廣播
3.花式索引指利用整數數組進行索引;花式索引跟切片不同，它老是將數據複製到新數組.