Python data analysis science library pandas (statistical analysis and decision)

One 、Pandas

• It is selected by NumPy Design for the foundation , In data analysis Pandas and NumPy These two modules are often used together .
  Pandas It is the preferred library for data processing and data analysis .
  Pandas It provides a large number of standard data models and functions and methods required for efficient operation of large data sets ,
  Is making Python One of the important factors to become an efficient and powerful data analysis tool .
  Use Pandas library , You need to import it and name it pd：

import pandas as pd

Two 、pandas data structure

• 1）Series, One dimensional array with label ;
  2）DatetimeIndex, The time series ;
  3）DataFrame, Two dimensional table structure with label and variable size ;
  4）Panel, Three dimensional array with label and variable size .

3、 ... and 、series object

• 1)Series：pandas An object that provides a dictionary structure similar to a one-dimensional array , Used to store a row or column of data .
  2)Series The internal structure of an object is composed of two interrelated arrays , Which is used to store data （ Instant value ） Yes. value Main array ,
  Each element of the main array has a tag associated with it （ Index ）, These tags are stored in another called Index In the array .
  3) If the index is not explicitly specified at the time of creation, it is automatically used from 0 Start with a nonnegative integer as the index .
  4)pd.Series(data=None, index=None, dtype=None, name=None, copy=False)

• data： The data that produces the sequence , It can be ndarray、list Or a dictionary
  index： catalog index , The number of index values must be the same as data The length of the data is the same . If you omit index Parameters , Its index 0, 1, 2, ….
  dtype： The data type of the sequence element , If you omit pandas Infer data types from data types .
  name： The name of the sequence
  copy： Copy the data , Generate a copy of the data . The default value is false
  

''' 1.Series Object creation （1） Create with list Series Series([ data 1, data 2,…],index=[ Indexes 1, Indexes 2,…]) If you omit index, The index is automatically added '''
import pandas as pd
slist1 = pd.Series([1, 2, 3, 4])
# Automatically add index , Index from 0 Start 
slist2 = pd.Series([1,2,3,4],index=['a','b','c','d'])
slist3 = pd.Series([1,2,3,4],[chr(i+65) for i in range(4)])
print(slist1,slist2,slist3,sep='\n')
#(2) Use range establish Series
#Series(range(strat,stop,step),index=[ Indexes 1, Indexes 2,…])
# Omit index, Automatically add index , Index from 0 Start 
srange1 = pd.Series(range(3))
srange2 = pd.Series(range(2,10,3))
srange3 = pd.Series(range(2,10,2),index=[i for i in range(2,10,2) ])
print(srange1,srange2,srange3,sep='\n')
#(3) Use numpy One dimensional array creation Series
import numpy as np
snumpy1 = pd.Series(np.array([1, 2, 3, 4]))
snumpy2 = pd.Series(np.arange(6,10))
snumpy3 = pd.Series(np.linspace(20,30,5))
#20-30 It is divided into 5 A floating point number 
print(snumpy1,snumpy2,snumpy3,sep='\n')
# Created by array Series when , Reference to the object .
aa = np.array([2,3,4,5])
aas = pd.Series(aa)
aas1 = pd.Series(aa,copy=True)
print(aa,aas,sep='\n')
aa[3] = 999
aas[0] = 1000
print(aa,aas,aas1,sep='\n')
#(4) Create with dictionary Series, The key of the dictionary is the index 
sdict1 = pd.Series( {
'Python Programming ':95,' data structure ':90,' Advanced mathematics ':88} )# Tuples 
sdict2 = pd.Series(dict(python Programming =92, Advanced mathematics =89, data structure =96)) # Convert to tuple -> Indexes 
print(sdict1,sdict2,sep='\n')
# 2.Series View the value and index of 
# utilize values and index You can see Series Values and labels 
print(sdict1.index)
print(sdict1.values)
print(slist2.index)
print(sdict2.values)
# 3. Access and modify by index or slice Series Value 
sdict1 = pd.Series({
'python Programming ':85,' data structure ':87,' Advanced mathematics ':78})
print('\n','sdict1 Raw data '.center(20, '='))
print(sdict1)
sdict1['python Programming ']=98
# How to modify data 
print('\n'," modify sdict1['python Programming ']=89 Post data ".center(30, '='))
print(sdict1)
sdict1[' College English ']=87 # If the index does not exist , Then add the data 
print('\n','sdict1 Data after adding '.center(20, '='))
print(sdict1)
# You can also use the default index for modification and access 
sdict1[0:2]=0 # Local modification in the form of slices 
sdict1[3]=999
print('\n','sdict1[0:2]=0 Slice modification sdict1[3]=999 Data after index modification '.center(40, '='),sdict1,sep='\n')
# 4.Series Operations between objects 
#（1） Arithmetic operation and mathematical function operation 
ss = pd.Series(np.linspace(-2,2,4))
print('\n','ss The original data '.center(14, '='),ss,sep='\n')
print('\n',' Yes ss Calculate the absolute value of all data , Retain 3 Decimal place '.center(20, '='),round(abs(ss),3),sep='\n')
print(np.sqrt(abs(ss)))
# round() Method returns a floating-point number x Round the value of . Its syntax :round( x [, n] )
# Parameters ：x -- Numerical expression ;n -- Numerical expression , From the decimal places .
# abs Function to find the absolute value 
print('\n','ss The arithmetic operation of '.center(14, '='))
print(' Add +5:',ss+5,' reduce -5:',ss-5,' ride *5',ss*5,' except /5',ss/5,sep='\n')
# （2） Select the elements that meet the criteria 
ss = pd.Series([np.random.randint(1,50) for i in range(10)])
# Screening is greater than 25 All the numbers of 
ss1 = ss[ss>25]
print('\n',ss,ss1,sep='\n')
# Screening can be 5 Divisible number 
ss2 = ss[ss%5==0]
print(' Screening can be 5 Divisible number \n',ss2)
# select 20 To 30 Number between 
print(' select 20 To 30 Number between '.center(20,'='))
ss3 = ss[(20<=ss) & (ss<=30)]
print(ss3)
# Screen out less than 20 And greater than 40 Number of numbers 
print(' Screen out less than 20 And greater than 40 Number of numbers '.center(20,'='))
ss3 = ss[(40<=ss) | (ss<=20)]
print(ss3)
# （3） Modify the index reset_index(drop=True)
ss_index1 = ss1.reset_index()
print(ss_index1,type(ss_index1))
ss_index2 = ss1.reset_index(drop=True)
# Delete the original index 
print(ss_index2,type(ss_index2))
# 5.Series Object statistical operation 
stu1 = pd.Series({
'python Programming ':85,' data structure ':87,' Advanced mathematics ':96,' College English ':76})
# Find the maximum index and the minimum index 
print('stu1 Index of maximum value '.center(20, '='))
print(stu1.idxmax())
print('stu1 Index of minimum value '.center(20, '='))
print(stu1.idxmin())
# Calculating mean mean()
print('\n mean value ：',stu1.mean())
print('stu1 Larger than average data in '.center(30,'='))
stu_mean=stu1[stu1>stu1.mean()]
print(stu_mean)
# Determine whether an element exists isin()
nn = [i for i in range(80,90)]
nn1 = stu1.isin(nn)
# Its return value is True or False
print(' With bool Value of output 80--89 Data between '.center(30, '='))
print(nn1)
print(' Only the output bool by True The data of '.center(30, '='))
print(nn1[nn1.values==True]) # The output value is True The data of 
nn1_index = nn1[nn1.values==True].index
print(' Output True Index of data '.center(30, '='))
print(nn1_index)
print(list(nn1_index))
# Project analysis 
import pandas as pd
num = pd.Series([np.random.randint(1,10) for i in range(20)])
print('\n primary Series\n',num)
# take Series Delete duplicate elements in , duplicate removal .
print(' take Series De duplication of elements in unique()'.center(30,'='))
print(num.unique())
# Count the number of occurrences of elements value_counts()
print(' Count the number of occurrences of elements value_counts()'.center(30,'='))
print(num.value_counts())
# Sort , Sort by value sort_values(ascending=True), Sort ascending by default , primary Series unchanged 
print(' Sort by value sort_values()'.center(30,'='))
print(num.sort_values())
# Sort by value sort_values(ascending=False), Sort in descending order , primary Series unchanged 
print(' Sort by value sort_values(ascending=False)'.center(30,'='))
print(num.sort_values(ascending=False))

The operation results are as follows ：

• 0 1
  1 2
  2 3
  3 4
  dtype: int64
  a 1
  b 2
  c 3
  d 4
  dtype: int64
  A 1
  B 2
  C 3
  D 4
  dtype: int64
  0 0
  1 1
  2 2
  dtype: int64
  0 2
  1 5
  2 8
  dtype: int64
  2 2
  4 4
  6 6
  8 8
  dtype: int64
  0 1
  1 2
  2 3
  3 4
  dtype: int32
  0 6
  1 7
  2 8
  3 9
  dtype: int32
  0 20.0
  1 22.5
  2 25.0
  3 27.5
  4 30.0
  dtype: float64
  [2 3 4 5]
  0 2
  1 3
  2 4
  3 5
  dtype: int32
  [1000 3 4 999]
  0 1000
  1 3
  2 4
  3 999
  dtype: int32
  0 2
  1 3
  2 4
  3 5
  dtype: int32
  Python Programming 95
  data structure 90
  Advanced mathematics 88
  dtype: int64
  python Programming 92
  Advanced mathematics 89
  data structure 96
  dtype: int64
  Index([‘Python Programming ’, ‘ data structure ’, ‘ Advanced mathematics ’], dtype=‘object’)
  [95 90 88]
  Index([‘a’, ‘b’, ‘c’, ‘d’], dtype=‘object’)
  [92 89 96]

• =sdict1 Raw data =
  python Programming 85
  data structure 87
  Advanced mathematics 78
  dtype: int64

• = modify sdict1[‘python Programming ’]=89 Post data =
  python Programming 98
  data structure 87
  Advanced mathematics 78
  dtype: int64

• sdict1 Data after adding =
  python Programming 98
  data structure 87
  Advanced mathematics 78
  College English 87
  dtype: int64

• =sdict1[0:2]=0 Slice modification sdict1[3]=999 Data after index modification =
  python Programming 0
  data structure 0
  Advanced mathematics 78
  College English 999
  dtype: int64

• ss The original data =
  0 -2.000000
  1 -0.666667
  2 0.666667
  3 2.000000
  dtype: float64

• = Yes ss Calculate the absolute value of all data , Retain 3 Decimal place =
  0 2.000
  1 0.667
  2 0.667
  3 2.000
  dtype: float64
  0 1.414214
  1 0.816497
  2 0.816497
  3 1.414214
  dtype: float64

• =ss The arithmetic operation of ==
  Add +5:
  0 3.000000
  1 4.333333
  2 5.666667
  3 7.000000
  dtype: float64
  reduce -5:
  0 -7.000000
  1 -5.666667
  2 -4.333333
  3 -3.000000
  dtype: float64
  ride *5
  0 -10.000000
  1 -3.333333
  2 3.333333
  3 10.000000
  dtype: float64
  except /5
  0 -0.400000
  1 -0.133333
  2 0.133333
  3 0.400000
  dtype: float64

• 0 28
  1 29
  2 33
  3 25
  4 21
  5 46
  6 17
  7 13
  8 13
  9 20
  dtype: int64
  0 28
  1 29
  2 33
  5 46
  dtype: int64
  Screening can be 5 Divisible number
  3 25
  9 20
  dtype: int64
  select 20 To 30 Number between
  0 28
  1 29
  3 25
  4 21
  9 20
  dtype: int64
  = Screen out less than 20 And greater than 40 Number of numbers =
  5 46
  6 17
  7 13
  8 13
  9 20
  dtype: int64
  index 0
  0 0 28
  1 1 29
  2 2 33
  3 5 46 <class ‘pandas.core.frame.DataFrame’>
  0 28
  1 29
  2 33
  3 46
  dtype: int64 <class ‘pandas.core.series.Series’>
  =stu1 Index of maximum value =
  Advanced mathematics
  =stu1 Index of minimum value =
  College English

• mean value ： 86.0
  stu1 Larger than average data in =
  data structure 87
  Advanced mathematics 96
  dtype: int64
  With bool Value of output 80–89 Data between
  python Programming True
  data structure True
  Advanced mathematics False
  College English False
  dtype: bool
  = Only the output bool by True The data of ==
  python Programming True
  data structure True
  dtype: bool
  = Output True Index of data =
  Index([‘python Programming ’, ‘ data structure ’], dtype=‘object’)
  [‘python Programming ’, ‘ data structure ’]

• primary Series
  0 4
  1 7
  2 4
  3 5
  4 1
  5 1
  6 5
  7 8
  8 6
  9 7
  10 4
  11 8
  12 9
  13 1
  14 6
  15 5
  16 6
  17 7
  18 2
  19 1
  dtype: int64
  take Series De duplication of elements in unique()=
  [4 7 5 1 8 6 9 2]
  = Count the number of occurrences of elements value_counts()=
  1 4
  7 3
  6 3
  5 3
  4 3
  8 2
  9 1
  2 1
  dtype: int64
  Sort by value sort_values()
  19 1
  13 1
  4 1
  5 1
  18 2
  10 4
  0 4
  2 4
  3 5
  15 5
  6 5
  8 6
  14 6
  16 6
  1 7
  17 7
  9 7
  7 8
  11 8
  12 9
  dtype: int64
  Sort by value sort_values(ascending=False)
  12 9
  11 8
  7 8
  9 7
  17 7
  1 7
  16 6
  14 6
  8 6
  6 5
  15 5
  3 5
  2 4
  0 4
  10 4
  18 2
  5 1
  4 1
  13 1
  19 1
  dtype: int64