Python——NumPy库：ndarray(多维数组对象)

目录

1、生成ndarray

2、切片和索引

2.1常规的切片索引

3、布尔索引

4、神奇索引

---

1、生成ndarray

简单举个例子：

# 一维数组
data1 = [4,5,6,2]
arr = np.array(data1)
arr
>>>array([4, 5, 6, 2])

# 二维数组
data2 = [[4,5,63,9],[5,63,9,2]]
arr1 = np.array(data2)
arr1
>>>array([[ 4,  5, 63,  9],
       [ 5, 63,  9,  2]])

下图是常规的创建数组的函数说明，创建方法与上文类似，这里不再陈述

---

2、切片和索引

2.1常规的切片索引

基本索引如下，内容简单，不过多陈述。

# 一维
arr3 = np.arange(10)
>>>array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
# 常规的索引
arr3[0]
arr3[4:6]

# 多维
arr4 = np.arange(20).reshape(2,2,5)
>>>array([[[ 0,  1,  2,  3,  4],
        [ 5,  6,  7,  8,  9]],

       [[10, 11, 12, 13, 14],
        [15, 16, 17, 18, 19]]])
# 索引与一维类似 arr4[0,1,2] 0是第0层，1是第一行，2是第二列
arr4[1,0] # 返回的是一维数组
>>>array([10, 11, 12, 13, 14])

下面的图可以很形象的展示基本索引:

基本切片

切片需要注意的是改变切片的数据，原数组的数据也会跟着改变。

arr3 = np.arange(10)
arr3[4:6] = 12
arr3
>>>array([ 0,  1,  2,  3, 12, 12,  6,  7,  8,  9]
# 如果改变切片的值，原数组也会改变
arr_s = arr3[4:6]
arr_s
>>>array([12, 12])

arr_s[0] = 8
arr3
>>>array([ 0,  1,  2,  3,  8, 12,  6,  7,  8,  9])

---

3、布尔索引

names = np.array(['a','b','c','d','a','c'])
data = np.random.randn(6,4)
names
>>> array(['a', 'b', 'c', 'd', 'a', 'c'], dtype='<U1')
data
>>>array([[-0.33682941,  0.462736  , -0.26948415,  1.06959446],
       [-0.3228036 ,  1.06947988, -0.27977893,  0.50433884],
       [ 0.05732912,  0.62538345,  0.55504195, -0.66662955],
       [ 0.32478243, -0.39012893,  2.06306486, -0.47683196],
       [-1.06078269,  1.8092996 ,  0.33705284,  0.24290218],
       [-0.62086775, -0.65675696,  0.57024677, -0.00484062]])


names == 'a'
>>> array([ True, False, False, False,  True, False])
 data[names == 'a']
>>>array([[-0.33682941,  0.462736  , -0.26948415,  1.06959446],
       [-1.06078269,  1.8092996 ,  0.33705284,  0.24290218]])

# 在此基础上进行索引：
data[names == 'a',2:]
>>>array([[-0.26948415,  1.06959446],
       [ 0.33705284,  0.24290218]])


# 取反操作： ~和！=结果是一样的
names !='a'
>>>array([False,  True,  True,  True, False,  True])

data[~(names == 'a')]
>>>array([[-0.3228036 ,  1.06947988, -0.27977893,  0.50433884],
       [ 0.05732912,  0.62538345,  0.55504195, -0.66662955],
       [ 0.32478243, -0.39012893,  2.06306486, -0.47683196],
       [-0.62086775, -0.65675696,  0.57024677, -0.00484062]])

data[names !='a']
>>>array([[-0.3228036 ,  1.06947988, -0.27977893,  0.50433884],
       [ 0.05732912,  0.62538345,  0.55504195, -0.66662955],
       [ 0.32478243, -0.39012893,  2.06306486, -0.47683196],
       [-0.62086775, -0.65675696,  0.57024677, -0.00484062]])

---

4、神奇索引

当有一个二维数组，为了选出一个特定顺序的子集，可以传递一个列表

arr = np.arange(32).reshape((8,4))
arr
>>>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]])
# 传递一个列表
arr[[1,5,7,2]]
>>>array([[ 4,  5,  6,  7],
       [20, 21, 22, 23],
       [28, 29, 30, 31],
       [ 8,  9, 10, 11]])

arr[[1,5,7,2],[0,3,1,2]]
>>>array([ 4, 23, 29, 10])
# 结果可以看出，传递一个列表是按照行进行输出，两个列表就是按照坐标（1，0）（5，3）...

# 如果要实现输出特定矩形区域
arr[[1,5,7,2]][:,[0,3,1,2]]
>>> array([[ 4,  7,  5,  6],
       [20, 23, 21, 22],
       [28, 31, 29, 30],
       [ 8, 11,  9, 10]])