python学习笔记15 模块numpy函数

Time：20181019

NumPy是Python语言的一个扩充程序库。支持高级大量的维度数组与矩阵运算，此外也针对数组运算提供大量的数学函数库。

1、np.newaxis：

np.newaxis:放在第几个位置，就会在shape里面看到相应的位置增加了一个维数
>>> import numpy as np
>>> x = np.random.randint(1, 8, size=5)
>>> x
array([7, 2, 3, 5, 2]) -
>>> x1 = x[np.newaxis, :]
>>> x1
array([[7, 2, 3, 5, 2]])-->1x5
>>> x2 = x[:, np.newaxis]
>>> x2
array([[7],
[2],
[3],
[5],
[2]])--->5x1
>>>
小结：多维数据提取一维数据后还原为多维数组

例子form:https://blog.csdn.net/xtingjie/article/details/72510834

1.1 x_data , y_label= produceData(10,6,-4,1000)

X_Col1 = np.random.uniform( r1*np.cos(theta1),r2*np.cos(theta1),num)[:, np.newaxis]

[[ 6.43537340e+00]
[ 6.82978402e+00]
[ 2.09355610e+00]
[-7.97087355e+00]
[-1.11236278e+01]
[-4.22396005e+00]
[-5.26542737e+00]
[-6.68204222e+00]
[ 8.44659647e+00]
[-3.28716816e+00]
[ 1.04212395e+01]
....

X_Col = np.random.uniform(r1 * np.cos(theta1), r2 * np.cos(theta1), num)

float64\

[ 4.00186233e+00 1.11634938e+01 1.60572965e+00 -7.76467945e+00
-1.17435194e+01 -5.57723441e+00 -6.79744608e+00 -1.07945486e+01
5.70181510e+00 -4.77393238e+00 7.95908523e+00 1.92325205e+00
-7.18504140e+00 -1.22510116e+01 2.55674780e+00 4.64671047e+00
-8.01048001e+00 1.08157827e+01 -1.94160498e+00 -6.17413987e+00
-1.11653391e+01 6.05140733e+00 7.07141974e+00 -6.10921790e+00
-3.80019914e+00 -1.02909642e+01 1.96867941e+00 5.24443249e+00
4.78028897e+00 1.19571914e+01 6.28946614e+00 4.79423174e+00
7.97336424e+00 -1.07988410e+00 1.12222201e+01 -6.46459059e+00
4.29360952e-01 9.53635479e+00 8.88714479e+00 -4.82616647e+00
-9.49782749e+00 8.26966741e+00 -8.47419249e+00 -1.06866070e+00
1.01734433e+00 -7.81109968e+00 -3.93336388e+00 -1.20250022e+01
9.40833887e+00 1.09078915e+01 6.83920543e+00 -1.16427596e+01
1.03922819e+01 -7.88152648e+00 -1.14732064e+01 -1.06747656e+01
8.80438740e+00 -8.43821282e+00 7.70779315e+00 -3.28956946e+00
.......

数据对不上：？？（见下）

----------------------------

X_Col1 =np.random.uniform( r1*np.cos(theta1),r2*np.cos(theta1),num)

X_Col1 =X_Col[:, np.newaxis]

float64
[ 9.63057718e+00 -8.89313282e+00 3.86946000e+00 4.95165355e-01
2.45909802e+00 -1.05677993e+00 6.11050453e+00 -5.50608221e+00
-8.02435169e+00 -9.48476236e+00 1.09073492e+01 -7.26462285e+00
1.08027047e+01 -1.25535557e+01 1.04945933e+01 -1.14269865e+01
-2.67661388e+00 -6.55422847e+00 -2.84605786e+00 5.59903410e+00
5.21495147e+00 9.19405019e+00 -1.08143459e+01 5.35552559e-01
......]

[[ 9.63057718e+00]
[-8.89313282e+00]
[ 3.86946000e+00]
[ 4.95165355e-01]
[ 2.45909802e+00]
[-1.05677993e+00]
[ 6.11050453e+00]
[-5.50608221e+00]
[-8.02435169e+00]
[-9.48476236e+00]
[ 1.09073492e+01]
..........]

把1x1000数据转换为1000x1数据

2、

numpy.vstack()函数

函数原型：numpy.vstack(tup)

等价于：np.concatenate(tup, axis=0） if tup contains arrays thatare at least 2-dimensional.

>>> a = np.array([1, 2, 3])
>>> b = np.array([2, 3, 4])
>>> np.vstack((a,b))
array([[1, 2, 3],
[2, 3, 4]])
>>>
>>> a = np.array([[1], [2], [3]])
>>> b = np.array([[2], [3], [4]])
>>> np.vstack((a,b))
array([[1],
[2],
[3],
[2],
[3],
[4]])

numpy.hstack()函数函数原型：numpy.hstack(tup) 按列合并

>>> a = np.array((1,2,3))
>>> b = np.array((2,3,4))
>>> np.hstack((a,b))
array([1, 2, 3, 2, 3, 4])
>>> a = np.array([[1],[2],[3]])
>>> b = np.array([[2],[3],[4]])
>>> np.hstack((a,b))
array([[1, 2],
[2, 3],
[3, 4]])

类似有：Python numpy函数hstack() vstack() stack() dstack() vsplit() concatenate()

3、transpose:

def produce_random_data(r,w,d,num):
    X1 = np.random.uniform(-r-w/2,2*r+w/2, num)
    X2 = np.random.uniform(-r - w / 2-d, r+w/2, num)
    X = np.vstack((X1, X2))
    print( X.transpose())

>>> a = array([[[ 0, 1, 2, 3],
[ 4, 5, 6, 7]],
[[ 8, 9, 10, 11],
[12, 13, 14, 15]]])
>>> b = a.transpose(１，０，２)
array([[[ 0, 1, 2, 3],
[ 8, 9, 10, 11]],
[[ 4, 5, 6, 7],
[12, 13, 14, 15]]])
原文：https://blog.csdn.net/zhangleaimeiling/article/details/78052235

数组a中10的坐标为a（1,0,3），经过transpose(1,0,2)转置后的数组ｂ中的10的坐标为b（0,1,3）。原始的transpose参数(默认的参数)为(0,1,2)，这个转置相当于将第一个坐标与第二坐标进行了互换。

4、tensorflow中的placeholder及用法。placeholder，中文意思是占位符，在tensorflow中类似于函数参数，运行时必须传入值。

python学习笔记15 模块numpy函数

猜你喜欢