2021-08-30 机器学习数据处理之numpy基础

import numpy as np

t11 = np.array([1,2,3,4])
type(t1)

numpy.ndarray

t22 = np.array(range(10))
print("t2=np.array(range(10)) = ",t2)
print("type(t2)=",type(t2))

t33 = np.arange(10)
print("t3=np.arange(10) = ",t3)
print("type(t3)=",type(t3))

t2=np.array(range(10)) =  [[4394029  320053    5931   46245]
 [7860119  185853   26679       0]
 [5845909  576597   39774  170708]
 ...
 [ 142463    4231     148     279]
 [2162240   41032    1384    4737]
 [ 515000   34727     195    4722]]
type(t2)= <class 'numpy.ndarray'>
t3=np.arange(10) =  [0 1 2 3 4 5 6 7 8 9]
type(t3)= <class 'numpy.ndarray'>

t22.reshape(2,5)

array([[0, 1, 2, 3, 4],
       [5, 6, 7, 8, 9]])

us_file_path = "./youtube_video_data/US_video_data_numbers.csv"
uk_file_path = "./youtube_video_data/GB_video_data_numbers.csv"

t1 = np.loadtxt(us_file_path,delimiter=",",dtype="int",unpack=True)
#unpack参数表示 矩阵的转置（把列向量转换成行向量）
t2 = np.loadtxt(us_file_path,delimiter=",",dtype="int")

print('t1:',t1,'\n')
print('t2',t2,'\n')

print("*"*20)

t1: [[4394029 7860119 5845909 ...  142463 2162240  515000]
 [ 320053  185853  576597 ...    4231   41032   34727]
 [   5931   26679   39774 ...     148    1384     195]
 [  46245       0  170708 ...     279    4737    4722]] 

t2 [[4394029  320053    5931   46245]
 [7860119  185853   26679       0]
 [5845909  576597   39774  170708]
 ...
 [ 142463    4231     148     279]
 [2162240   41032    1384    4737]
 [ 515000   34727     195    4722]] 

********************

t22+1

array([ 1,  2,  3,  4,  5,  6,  7,  8,  9, 10])

t22**2

array([ 0,  1,  4,  9, 16, 25, 36, 49, 64, 81])

t=np.arange(24).reshape(4,6)
t

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]])

t<12

array([[ True,  True,  True,  True,  True,  True],
       [ True,  True,  True,  True,  True,  True],
       [False, False, False, False, False, False],
       [False, False, False, False, False, False]])

t[t<12]=1
t

array([[ 1,  1,  1,  1,  1,  1],
       [ 1,  1,  1,  1,  1,  1],
       [12, 13, 14, 15, 16, 17],
       [18, 19, 20, 21, 22, 23]])

np.where(t<12,0,12)

array([[ 0,  0,  0,  0,  0,  0],
       [ 0,  0,  0,  0,  0,  0],
       [12, 12, 12, 12, 12, 12],
       [12, 12, 12, 12, 12, 12]])

t

array([[ 1,  1,  1,  1,  1,  1],
       [ 1,  1,  1,  1,  1,  1],
       [12, 13, 14, 15, 16, 17],
       [18, 19, 20, 21, 22, 23]])

#.clip(a,b)：裁剪功能，将矩阵中小于a的全部赋值成a,大于b的全部赋值为b，一般a≤b，但实际上b可以小于a
t.clip(12,19)

array([[12, 12, 12, 12, 12, 12],
       [12, 12, 12, 12, 12, 12],
       [12, 13, 14, 15, 16, 17],
       [18, 19, 19, 19, 19, 19]])

t[3,3:6]=9999
t

array([[   1,    1,    1,    1,    1,    1],
       [   1,    1,    1,    1,    1,    1],
       [  12,   13,   14,   15,   16,   17],
       [  18,   19,   20, 9999, 9999, 9999]])

t

array([[   1,    1,    1,    1,    1,    1],
       [   1,    1,    1,    1,    1,    1],
       [  12,   13,   14,   15,   16,   17],
       [  18,   19,   20, 9999, 9999, 9999]])

t.sum()

30153

t.sum(axis=0)

array([   32,    34,    36, 10016, 10017, 10018])

t.sum(0)

array([   32,    34,    36, 10016, 10017, 10018])

t.sum(axis=1)

array([    6,     6,    87, 30054])

t.sum(1)

array([    6,     6,    87, 30054])

• 中值mean 和 平均值median没有必然的关系。
• 中值是将所给的一组数从小到大或从大到小排列，奇数个数的话取中间的数字，偶数个数的话取中间两个数的平均数；而平均值就是把这组数相加，然后除以这组数的个数。
• 中值的优点是不受偏大或偏小数据的影响，很多情况下用它代表全体数据的一般水平更合适。如果数列中存在极端变量值，用中位数做代表值就比平均数更好。

#均值
t.mean()

1256.375

t.mean(0)#小括号内简写了： axis=0

array([   8.  ,    8.5 ,    9.  , 2504.  , 2504.25, 2504.5 ])

t.mean(1)

array([1.000e+00, 1.000e+00, 1.450e+01, 5.009e+03])

#中值
np.median(t)

6.5

np.median(t,axis=0)

array([6.5, 7. , 7.5, 8. , 8.5, 9. ])

np.median(t,axis=1)

array([1.0000e+00, 1.0000e+00, 1.4500e+01, 5.0095e+03])

t

array([[   1,    1,    1,    1,    1,    1],
       [   1,    1,    1,    1,    1,    1],
       [  12,   13,   14,   15,   16,   17],
       [  18,   19,   20, 9999, 9999, 9999]])

#最大值
t.max()

9999

t.max(axis=0)

array([  18,   19,   20, 9999, 9999, 9999])

t.min()

1

t.min(axis=0)

array([1, 1, 1, 1, 1, 1])

#极值:最大值和最小值的差值
np.ptp(t)

9998

np.ptp(t,axis=0)

array([  17,   18,   19, 9998, 9998, 9998])

#标准差
t.std()

3304.409324580567

t.std(axis=0)

array([   7.31436942,    7.79422863,    8.27647268, 4327.24404211,
       4327.10026317, 4326.95652278])

t.std(axis=1)

array([0.00000000e+00, 0.00000000e+00, 1.70782513e+00, 4.99000003e+03])

---

import numpy as np
from matplotlib import  pyplot as plt

# us_file_path = "./youtube_video_data/US_video_data_numbers.csv"
uk_file_path = "./youtube_video_data/GB_video_data_numbers.csv"

# t1 = np.loadtxt(us_file_path,delimiter=",",dtype="int",unpack=True)
t_uk = np.loadtxt(uk_file_path,delimiter=",",dtype="int")

print('最初的t_uk:',t_uk)
print('t_uk[:,1]<=500000:',t_uk[:,1]<=500000)
print('t_uk[t_uk[:,1]<=500000]:',t_uk[t_uk[:,1]<=500000])

#选择喜欢书比50万小的数据
t_uk = t_uk[t_uk[:,1]<=500000]
print('t_uk:',t_uk)

t_uk_comment = t_uk[:,-1]
t_uk_like = t_uk[:,1]

plt.figure(figsize=(20,8),dpi=80)
plt.scatter(t_uk_like,t_uk_comment)

plt.show()

最初的t_uk: [[7426393  782040   13548     705]
 [ 494203    2651    1309       0]
 [ 142819   13119     151    1141]
 ...
 [ 109222    4840      35     212]
 [ 626223   22962     532    1559]
 [  99228    1699      23     135]]
t_uk[:,1]<=500000: [False  True  True ...  True  True  True]
t_uk[t_uk[:,1]<=500000]: [[ 494203    2651    1309       0]
 [ 142819   13119     151    1141]
 [1580028   65729    1529    3598]
 ...
 [ 109222    4840      35     212]
 [ 626223   22962     532    1559]
 [  99228    1699      23     135]]
t_uk: [[ 494203    2651    1309       0]
 [ 142819   13119     151    1141]
 [1580028   65729    1529    3598]
 ...
 [ 109222    4840      35     212]
 [ 626223   22962     532    1559]
 [  99228    1699      23     135]]

数据拼接

t33=np.arange(12).reshape(2,6)
t44=np.arange(12,24).reshape(2,6)
print('t33:\n',t33,'\n\n','t44:\n',t44)

t33:
 [[ 0  1  2  3  4  5]
 [ 6  7  8  9 10 11]] 

 t44:
 [[12 13 14 15 16 17]
 [18 19 20 21 22 23]]

#竖直拼接
np.vstack((t33,t44))

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]])

#横向拼接
np.hstack((t33,t44))

array([[ 0,  1,  2,  3,  4,  5, 12, 13, 14, 15, 16, 17],
       [ 6,  7,  8,  9, 10, 11, 18, 19, 20, 21, 22, 23]])

t

array([[   1,    1,    1,    1,    1,    1],
       [  16,   13,   15,   12,   14,   17],
       [   1,    1,    1,    1,    1,    1],
       [9999,   19, 9999,   18,   20, 9999]])

#行交换
t[[1,2],:]=t[[2,1],:]
t

array([[   1,    1,    1,    1,    1,    1],
       [   1,    1,    1,    1,    1,    1],
       [  16,   13,   15,   12,   14,   17],
       [9999,   19, 9999,   18,   20, 9999]])

#列交换
t[:,[0,4]]=t[:,[4,0]]
t

array([[   1,    1,    1,    1,    1,    1],
       [   1,    1,    1,    1,    1,    1],
       [  14,   13,   15,   12,   16,   17],
       [  20,   19, 9999,   18, 9999, 9999]])

#numpy获取最大值最小值的下标
np.argmax(t)#第一个最大值的位置

20

np.argmax(t,axis=0)

array([3, 3, 3, 3, 3, 3])

np.argmax(t,axis=1)

array([0, 0, 5, 2])

np.argmin(t,axis=1)

array([0, 0, 3, 3])

# 创建3×4全0的数组: 
np.zeros((3,4))

array([[0., 0., 0., 0.],
       [0., 0., 0., 0.],
       [0., 0., 0., 0.]])

# 创建3×4全1的数组: 
np.ones((3,4))

array([[1., 1., 1., 1.],
       [1., 1., 1., 1.],
       [1., 1., 1., 1.]])

# 创建一个对角线为1的3x3正方形数组(方阵)：
np.eye(3)

array([[1., 0., 0.],
       [0., 1., 0.],
       [0., 0., 1.]])

np.random.rand(2,3,2)

array([[[0.35950567, 0.08177032],
        [0.89808166, 0.73285583],
        [0.23462587, 0.46792952]],

       [[0.02080821, 0.55948812],
        [0.25205097, 0.75559711],
        [0.93791326, 0.24709769]]])