numpy and pandas学习记录

numpy_learn

# -*- coding:UTF-8 -*-
# @time: 2022-01-19 14:38
# @file: numpy_learn.py
# @software: PyCharm
# author CJH

import numpy as np

# 创建数组，再转化为矩阵
array=np.array([[1,2,3],
                [4,5,6]])
print(array)
print('number of dim:',array.ndim)  #二维数组
print('shape:',array.shape)         #输出矩阵的行列数
print('array_size:',array.size)     #矩阵的大小

a = np.array([[2,23,4],
             [3,33,34]],dtype=np.double)    #设置浮点数，float
print(a.dtype)

b=np.ones((4,5),dtype=np.double)     #生成3,4纬的矩阵
print(b)

a1=np.arange(12).reshape(3,4)      #生成1,2....11数列
print(a1)

a2=np.linspace(1,10,20).reshape(4,5)      #等差数列生成20，1~10
print(a2)

b_a2=b-a2       #逐个元素相减
print(b_a2)

b_a22=b_a2**2
print(b_a22)

c=10*np.sin(b_a22)     #三角函数，大小判断输出
print(c<5)

a_mut=np.array([[1,1],
                [2,2]],dtype=np.double)
b_mut=np.arange(4).reshape((2,2))
# 运行矩阵乘法
cmul = a_mut*b_mut   #逐个相乘
cdot = np.dot(a_mut,b_mut)    #矩阵乘法  c_dot = a_mut.dot(b_mut)
print(cmul)
print(cdot)

a_random = np.random.random((2,4))*10   #随机生成2,4的矩阵
print(a_random)

print(np.sum(a_random,axis=0))
print(np.max(a_random,axis=0))
print(np.min(a_random,axis=1))   #axis=1,求行的。。得出列向量，axis=0求列的。。。得出行向量

# 寻找最大值的索引
# array=np.array([[1,2,3],
#                 [4,5,6]])
print(np.argmax(array,axis=1))
print(np.max(array,axis=1))    #加上arg求索引，不加arg求数值
#python中索引使用[]中括号
print(np.mean(array,axis=1))    #mean求平均数
print(array.mean(axis=1))
print(np.median(array,axis=1))   # 求中位数

print(np.cumsum(array,axis=0))    #前缀和  ，0列求和，1行求和
print(np.diff(array,axis=1))      #前项差
print(array.T)                    #矩阵转置   np.transpose(array)
print(array.dot(array.T))         #矩阵乘法   未重新赋值，不影响原本矩阵的数值

print(np.clip(array,2,7))         #截数字，小于2的变为2，大于7的变为7

array1=np.arange(3,15).reshape(3,4)
print(array1)
print(array1[1][2])               #[1][2]=[1,2]

print(array1[1][:])               #:打印全部行or全部列   [1,:]
for i in array1:
    print(i)    #迭代输出每一个，先输出行，如要先输出列，使用array1.T,将其转置再迭代
    #python先遍历行

A = np.array([[1,1,1]])
B = np.array([[2,2,2]])           #列不相等，无法进行列合并
print(np.hstack((A,B)))         #行合并
print(np.vstack((A,B)))         #列合并
print(A[np.newaxis,:].T)          #在行上加了一个维度，方便转置为列向量,再对行向量进行转置,
print(B.shape[0])                 #如果在一开始就加入2个[]就不需要在后期加维度
print(B.shape[1])
print(A.T)
C = np.concatenate((A,B,B,A),axis=0)      #任意指定在行还是列合并
print(C)

# 分割矩阵，二维数组
A_1=np.arange(12).reshape((3,4))
print(A_1)
print(np.split(A_1,2,axis=1))
#C_1=A_1[:,0][np.newaxis,:].T*1          #在赋值后加上*1，使得python重新创建一个变量
C_1 = A_1.copy()[:,0]
print(C_1)                              #在矩阵后加上[np.newaxis,:]可以扩充向量为二维向量
A_1[0,0]=1
print(C_1)

# pandas_learn

# -*- coding:UTF-8 -*-
# @time: 2022-01-21 11:17
# @file: pd_learn.py
# @software: PyCharm
# author CJH

import pandas as pd
import numpy as np

s = pd.Series([1,3,6,np.nan,44,10])
print(s)

dates = pd.date_range('20160101',periods=6)
print(dates)

df = pd.DataFrame(np.arange(24).reshape(6,4),index=dates,columns=['a','b','c','d'])
print(df)

df1=pd.DataFrame(np.arange(12).reshape(3,4))
print(df1)
#DataFrame转化为二维矩阵，并且可以加标签

print(df.describe())

print(df.sort_index(axis=0,ascending=False))

print(df['a'],df.a)   #选取某一列
print(df[0:3],df['20160101':'20160104'])   #选择行
print(df.loc[:,['a','b']])
df.iloc[2,2]=1111           #根据编号定位
print(df)

df.a[df.a>4]=0
print(df)

print(df.dropna(axis=0,how='any'))   #丢掉包括nan的行，how=all全部是nan才丢弃
df.iloc[0,0]=np.nan
print(df)
#print(df.fillna(value=100))             #将nan替换
print(df.isnull())                        #判断是否缺失数据
print((np.any(df.isnull())==True))          #判断是否丢失数据

## important pandas导入导出

data_1 = pd.read_csv('student.csv')
print(data_1)
print(data_1.iloc[0,0])
data_1.to_pickle('student.pickle')          #保存成pickle文件


## pandas 合并列表
# concatenating

data1=pd.DataFrame(np.ones((3,4))*0,columns=['a','b','c','d'])
data2=pd.DataFrame(np.ones((3,4))*1,columns=['f','b','c','d'])
data3=pd.DataFrame(np.ones((3,4))*2,columns=['a','b','c','d'])
#print(data1,'\n',data2)
res = pd.concat([data1,data2,data3],axis=0,ignore_index=True)     #axis=0竖向的合并，忽略列项标签
print(res)
res2 = pd.concat([data1,data2],join='inner',ignore_index=True)      #列方向合并，inner合并都有，outer全部保留
print(res2)
res1 = pd.concat([data1,data2],axis=1) #横向合并
print(res1)

res3=data1.append(data3,ignore_index=True)
print(res3)
s1 = pd.Series([1,2,3,4],index=['a','b','c','d'])
res4=res3.append(s1,ignore_index=True)          #合并
print(res4)

# merge  索引，key对照的合并

left = pd.DataFrame({'Key':['K0','K1','K2','K3'],
                     'A':['A0','A1','A2','A3'],
                     'B':['B0','B1','B2','B3']})
right = pd.DataFrame({'Key':['K0','K1','K2','K4'],
                      'A':['C0','C1','C2','C3'],
                      'D':['D0','D1','D2','D3']})
## 利用merge合并left和right
res5 = pd.merge(left,right,on=['Key'],how='inner')  #利用Key标签合并on = ['key1','key2']可以有多个key
# 默认使用inner合并，只合并相同的部分
print(res5)

res6 = pd.merge(left,right,on='Key',suffixes=['_left','_right'],how='outer')
print(res6)


## pandas 画图

import matplotlib.pyplot as plt
# plot data
#Series
# data =pd.Series(np.random.randn(1000),index=np.arange(1000))
# data = data.cumsum()

#DataFrame
data = pd.DataFrame(np.random.randn(1000,4),index=np.arange(1000),columns=list("ABCD"))
data =data.cumsum()
print(data.head(3))
# 打印scatter
# 'bar','hist','box','kde',
data.plot()
plt.show()

# 绘制散点图
ax = data.plot.scatter(x='A',y='B',color='DarkBlue',label='Class 1')
data.plot.scatter(x='A',y='C',color='DarkGreen',label='Class 2',ax=ax)  # 将多个线条绘制到同一个图上
data.plot.scatter(x='A',y='D',color='Black',label='Class 3',ax=ax)

plt.show()