Python-numpy

@[目录]

用于大型多维数组的数值计算

#!usr/bin/env python
# -*- coding:utf-8 -*-
"""
@author: lyf
@file: Task001.py
@time: 2021/10/28
@desc: numpy的创建与类型
"""

import numpy as np
import random

# 数组的几种定义方式
t1 = np.array([1,2,3])
print(t1)
print(t1.dtype)

t2 = np.array(range(10))
print(t2)
print(t2.dtype)

t3 = np.arange(10)
print(t3)
print(t3.dtype)

# 定义数组时，指定数组的类型
t4 = np.array([1,2,3],dtype='i1')
print(t4)
print(t4.dtype)

t5 = np.array([1,1,0,0,1,1,0],dtype='bool')
print(t5)
print(t5.dtype)

# 调整数据类型
t6 = t5.astype(dtype='i1')
print(t6)
print(t6.dtype)

# numpy中的小数
t7 = np.array([random.random() for i in range(10)])
print(t7)
print(t7.dtype)

t8 = np.round(t7,3)
print(t8)

import numpy as np

#查看数组的形状：shape
t1 = np.array([[1,2,3],[4,5,6]])
print('t1')
print(t1)
print(t1.shape)

print('*'*50)
t2 = np.array([[[1,2],[3,4]],[5,6],[[7,8,9],[10,11]]])
print('t2')
print(t2)
print(t2.shape)


#修改数组形状，二维数组更改为一维数组： reshape(1,n), t.shape[0]*t.shape[1],flatten
print('*'*50)
t3 = t1.reshape(1,6)
print('t3')
print(t3)
print(t3.shape)

print('*'*50)
t4 = t1.reshape(t1.shape[0]*t1.shape[1],)
print('t4')
print(t4)
print(t4.shape)

print("*"*50)
t5 = t1.flatten()
print('t5')
print(t5)

# 数组和数的四则运算：广播机制
print("*"*50)
t6 = t1 +2
print('t6')
print(t6)

#数组和数组的四则运算：数组行列一样时，对应位置数字相加
print("*"*50)
t7 = np.arange(6).reshape(2,3)
print('t7')
print(t7)
t8 = t7+t1
print('t8')
print(t8)

#数组和数组行列不同：广播机制，
print('*'*50)
t9 = np.arange(3)
print('t1')
print(t1)
print('t9')
print(t9)
t10 = t1+t9
print('t10')
print(t10)

np.loadtxt(frame，dtype=np.float,delimiter=None,skiprows=0,usecols=None,unpack=False)

import numpy as np

#先导入文件地址
uk_file_path = './gb_videos_data_numbers.csv'

print('*'*50)
t1 = np.loadtxt(uk_file_path,delimiter=',')
print('t1')
print(t1)

print('*'*50)
t2 = np.loadtxt(uk_file_path,delimiter=',',unpack=1)
#unpack实现转置功能
print('t2')
print(t2)

# numpy中的转置:t.transpose(),t.T,t2.swapaxes(1,0)交换轴
print('*'*50)
t3 = t1.transpose()
print('t3')
print(t3)
t4 = t1.T
print('t4')
print(t4)
t5 = t1.swapaxes(1,0)
print('t5')
print(t5)

import numpy as np

#先导入文件地址
uk_file_path = './gb_videos_data_numbers.csv'
t1 = np.loadtxt(uk_file_path,delimiter=',')
print('t1')
print(t1)

# 取列表中行
print('t1中的第三行')
print(t1[2])  #这两个语句同意print(t1[2,:])

print('t1中第三行及其以后的行')
print(t1[2:]) #print(t1[2:,:])

print('t1中不连续的多行，第3,5,8行')
print(t1[[2,4,7]]) #print(t1[[2,4,7],:])

#取列表中的列
print('取t1中第一列')
print(t1[:,0])
print('取连续多列')
print(t1[:,3:])
print('取不连续的多列')
print(t1[:,[1,3]])

# 取多行和多列
print('取第三行第四列')
print(t1[2,3])
print('取连续的多行和多列')
print(t1[2:6,1:3]) #取行列交叉点的值
print('取不连续的多行和多列')
print(t1[[1,7,3],[0,2,0]]) #取第一行第一列，第八行第三列，第四行第一列这三个值

直接赋值即可

import numpy as np
t = np.array([[2,3,4],[3,4,6],[77,8,6],[2,3,4]])

print('更改整行的值')
t[1]=1
print(t)

print('更改指定大小的值')
t[t>5] = 10
print(t)

# where的使用（相当于if else）
print('测试where的使用')
t1 = np.where(t<3,0,99)
print(t1)


# clip裁剪 t.clip(a,b)小于a替换为a,大于b替换为b
t2 = t.clip(3,9)
print(t2)

#将某处的值赋值为nan，nan为浮点类型，将原数据类型更改为浮点型之后才可赋值nan
t = t.astype(float)
t[2,1] = np.nan
print(t)

import numpy as np
t = np.array([[1,2,3,4],[3,4,5,6],[2,4,np.nan,6],[np.nan,8,5,3]])


t[1] = 0
print('t')
print(t)
print('统计数组中0的个数')
print(np.count_nonzero(t))

print('统计数组中nan的个数')
print(np.count_nonzero(np.isnan(t)))

print('行求和')
print(np.sum(t,axis=0))
print('列求和')
print(np.sum(t,axis=1))

import numpy as np

def fill_ndarray(t1):
    for i in range(t1.shape[1]):
        temp_col = t1[:,i]
        if np.count_nonzero(np.isnan(temp_col)) != 0:
            temp_col_not_nan = temp_col[temp_col == temp_col]
            temp_col[temp_col != temp_col] = temp_col_not_nan.mean()
    return t1

print('将数列中的nan替换为均值，或者中值')
if __name__ == '__main__':
    t1 = np.arange(12).reshape((3, 4)).astype(float)
    print('t1原始矩阵')
    print(t1)
    t1[1, 2:] = np.nan
    print('t1赋值nan后的矩阵')
    t2 = fill_ndarray(t1)
    print('t2')
    print(t2)

import numpy as np
from matplotlib import pyplot as plt

uk_file_path = './gb_videos_data_numbers.csv'
t_uk = np.loadtxt(uk_file_path,delimiter=',',dtype='int')

#取评论的数据
t_uk_comments = t_uk[:,-1]
t_uk_comments = t_uk_comments[t_uk_comments<=10000]

#选择比10000小的数据

d = 500
bin_nums = (t_uk_comments.max()-t_uk_comments.min())//d

plt.figure(figsize=(20,8),dpi=80)
plt.hist(t_uk_comments,bin_nums)

plt.show()

import numpy as np
from matplotlib import pyplot as plt

uk_file_path = './gb_videos_data_numbers.csv'
t_uk = np.loadtxt(uk_file_path,delimiter=',',dtype='int')

t_uk = t_uk[t_uk[:,1] < 20000]
t_uk_comments = t_uk[:,-1]
t_uk_like = t_uk[:,1]



plt.scatter(t_uk_comments,t_uk_like)
plt.show()

np.vstack竖直拼接,np.hstack水平拼接,

import numpy as np

t1 = np.array([[1,2,3],[2,3,4]])
t2 = np.array([[2,6,4],[2,9,1]])
print('t1')
print(t1)
print('t2')
print(t2)
print('拼接')
t3 = np.vstack((t1,t2))
t4 = np.hstack((t1,t2))
print('t3')
print(t3)
print('t4')
print(t4)


print('行交换')
t5 = t2[[1,0],:]
print(t5)

print('列交换')
t6 = t2[:,[1,0]]
print(t6)

import numpy as np

t1 = np.ones((3,2))

t2 = np.zeros(4,3)

np.eye(10)

np.argmax(t1,axis=1)

np.argmin(t2,axis=0)

print('numpy生成随机数')
np.round()
np.roundn()