Numpy numerical calculation basis

Catalog

  One ,Numpy The concept of

Two , Array

1, Properties of array

2, Array creation

2.1 array function

2.2 arrange function

2.3 linspace function

2.4 logspace function

2.5 zeros function

2.6 ones function

 2.7 eye function

2.8 diag function

3, Conversion of array data type

3.1,np. data type ( Data objects )

​ 3.2, Data objects .astype(np. data type )

​4, Array processing

 4.1, Index and slice of arrays

4.2, Modify array shape

4.3, Flattening and linking of arrays

4.4, Processing of array elements

4.5, Operation of array

5, Random number module

5.1,random function ： Generate [0,1) The random number in

5.2,rand function ： Generate uniformly distributed [0,1) The random number in

5.3,randn function ： Generating random numbers that obey normal distribution

5.4,randint function ： Generate random numbers with given upper and lower bounds

5.5,choice function ： Random sampling from a list or array

5.6,random The common random number generation function of the module

3、 ... and , matrix

1, Create a matrix

2, The vector product of a matrix

3, The quantity product of a matrix

Four , Read and write files

1,save Function is to save an array of data in binary format

2,savez Function can save multiple arrays to a file

3,load The function reads data from a binary file

4,savetxt() Function to write an array to a delimited text file .

5,loadtxt() Function to load a file into a two-dimensional array .

5、 ... and ,numpy Statistical analysis in

1,sort() Direct sort

2, Indirect sort

3, Repetition and de duplication

---

  One ,Numpy The concept of

Numpy It is a basic module for data scientific computing , It can not only complete the task of scientific calculation , And can be used as an efficient multidimensional data container , Can be used to store and process large matrices ,Numpy Data container can store any type of data , This makes Numpy Can quickly integrate various data .

Two , Array

1, Properties of array

attribute	explain
ndim	return int, Represents the dimension of the array
shape	return tuple. Represents the size of an array , about n That's ok m Columns of the matrix , Shape is (n,m)
size	return int, Represents the total number of array elements , Equal to the product of array shape
dtype	return data-type, Describes the type of elements in an array
itemsize	return int, Represents the size of each element of the array （ In bytes ）

2, Array creation

2.1 array function

Used to create one-dimensional or multi-dimensional arrays .np.array(object,dtype,ndim)

Parameter name	explain
object	Used to receive arrays , Represents the array you want to create , No default
dtype	receive data-type, Represents the data type required by the array , The default is None
ndim	receive int. Specifies the minimum dimension that the generated array should have , The default is None

# np.array Generate one dimensional array 
arr1=np.array([1,2,3,4])

# np.array Generate a two-dimensional array 
arr2=np.array([[1,2,3,4],[4,5,6,7]])

2.2 arrange function

Be similar to Python Self contained function range, By specifying the starting value 、 Final value and step size to create a one-dimensional arithmetic array . The created array does not contain a final value . The default step size is 1.

# np.arrange Create array 
arr=np.arange(1,10)

2.3 linspace function

By specifying the starting value 、 Final value and number of elements to create a one-dimensional arithmetic array . The created array contains the final value .

#  adopt np.linspace The creation start value is 1, The final value is 10, The number is 5 One dimensional arithmetic array of .
arr=np.linspace(1,10,5)

2.4 logspace function

And linspace Function similar to , But what it creates is an isometric sequence

#  adopt np.logspace The creation start value is 1, The final value is 10, The number is 5 One dimensional proportional array of .
arr=np.logspace(1,10,5)

2.5 zeros function

Creates an all zero array of the specified length or shape

 #  Create a length of 2 One dimensional all zero array 
arr=np.zeros(2)

#  Create an all zero array of three rows and four columns 
arr=np.zeros((3,4))

2.6 ones function

Creates an entire array of the specified length or shape

#  Create a length of 2 One dimensional full 1 Array 
arr=np.ones(2)

#  Create three rows and four columns of full 1 Array 
arr=np.ones((3,4))

 2.7 eye function

The elements used to create the main diagonal of the build are 1, The other elements are 0 Array of . Similar identity matrix

#  Create the elements on the main diagonal of four rows and four columns as 1, The other elements are 0 Array of 
arr=np.eye(4)

2.8 diag function

Create diagonal matrix , That is, all the elements except the diagonal are 0.

#  Create diagonal elements as 1 and 3, The other elements are 0 Diagonal matrix of .
arr=np.diag([1,3])

3, Conversion of array data type

3.1,np. data type ( Data objects )

Create an array and see its data type

arr3=np.arange(1,10)
arr3.dtype

adopt  np. data type ( Data objects ) The way to change its data type .

arr3=np.int64(arr3)

  The screenshot of the code running is as follows ：

 3.2, Data objects .astype(np. data type )

arr3.astype(np.int32)

The screenshot of the code running is as follows ：

4, Array processing

 4.1, Index and slice of arrays

4.1.1 One dimensional array index ： Array name [m:n], among m,n Subscript array elements

arr=np.arange(1,10)
arr[0:2]

The operation results are as follows  ：

 4.1.2 One dimensional array slices ： Array name [start,end,step],( Semi open interval )

arr[0:3:1]

The operation results are as follows ：

 4.1.3 2d array slice ： Array name [ Row index , Column index ], Semi open interval

Create a 2D array , The truncated row index is equal to 0 An array consisting of all the following rows and all the columns after the first column .

arr=np.array([[1,2,3,4],[4,5,6,7]])
arr[0:,1:]

The operation results are as follows ：

 4.1.4 Integer index ： Take two integers from the corresponding positions of the two sequences to form row subscript and column subscript .

4.1.5 Boolean index ： When the result object is a Boolean operation （ For example, the comparison operator ） When the result is , Use Boolean index

4.2, Modify array shape

4.2.1, Array name .reshape=( That's ok , Column ). Do not change the shape of the original array

# Create an array 
arr=np.arange(12)

arr.reshape(3,4)

4.2.2, Array name .resize=( That's ok , Column ). Do not change the shape of the original array , Returns a new array of the specified size .

arr.resize(4,3)

4.2.3 Set dimension f How to modify the shape of an array ：

arr.shape=(3,4)

The results are shown below ：

4.3, Flattening and linking of arrays

4.3.1 ravel() function ： You can flatten a multidimensional array into a one-dimensional array , primary arr unchanged . Flatten by row

#ravel() Function to flatten by line 
arr3=arr1.ravel()

The operation results are as follows ：

 4.3.2 flatten() function ： You can flatten a multidimensional array into a one-dimensional array , And you can choose to flatten by row or column

#flatten() The function flattens by line by default 
arr3=arr1.flatten()
#"F" Indicates flattening by column 
arr3=arr1.flatten("F")

The operation results are as follows ：

4.3.3 hstack() function , Left and right splicing

np.hstack((arr1,arr2))

The running results are as follows  

 4.3.4 vstack() function ： Upper and lower splicing

np.vstack((arr1,arr2))

The operation results are as follows ： 

 4.3.5 concatenate() function ： Used to connect two or more arrays of the same shape along the specified axis .

#axis = 1 Left and right splicing 
np.concatenate((arr1,arr2),axis = 1)

#axis = 0 Upper and lower splicing 
np.concatenate((arr1,arr2),axis = 0)

The operation results are as follows ： 

 4.3.6 spilt() function ： Divide an array into subarrays along a specific axis

#axis = 1  Split column 
np.split(arr1,3,axis=0)

#axis = 0  Split line 
np.split(arr1,3,axis=1)

The operation results are as follows ： 

4.3.7 transpose() function ： Realize the transpose of array , primary arr unchanged

# Method 1 
np.transpose(arr1)


# Method 2 
arr1.T

The operation results are as follows ： 

4.4, Processing of array elements

4.4.1  append() function ： Is to add a function at the end of the array

# Add by line 
np.append(arr,[[0, 0, 0]],axis=0)

# Add... By column 
np.append(arr,[[0],[0], [0]],axis=1)

The operation results are as follows ： 

4.4.2 insert() function ： Before a given index , Inserts elements in the input array along a given axis

# Press the line 
np.insert(arr,3,[[0,0,0],[1,1,1],[2,2,2]],axis=0)


# By column 
np.insert(arr,3,[[0,0,0],[1,1,1],[2,2,2]],axis=1)

give the result as follows ： 

4.4.3 delete() function ： Returns a new array that removes the specified subarray from the input array

# Delete the second line 
np.delete(arr,1,axis=0)

# Use slices to represent element ranges 
np.delete(arr,np.s_[1,1,1])

The operation results are as follows ：

4.5, Operation of array

Arrays of the same shape are calculated element by element at the element level , And arrays of different shapes , Then it is calculated according to the broadcast mechanism

5, Random number module

5.1,random function ： Generate [0,1) The random number in

5.2,rand function ： Generate uniformly distributed [0,1) The random number in

  

5.3,randn function ： Generating random numbers that obey normal distribution

  

5.4,randint function ： Generate random numbers with given upper and lower bounds

5.5,choice function ： Random sampling from a list or array

5.6,random The common random number generation function of the module

Function name	purpose
seed	Identify random seeds
permutation	Returns the random ordering of a sequence
shuffle	Randomly sort a sequence （ Change the original sequence ）

3、 ... and , matrix

1, Create a matrix

2, The vector product of a matrix

The operator uses *, And * The number of columns of the matrix on the left should be equal to the number of rows on the right , The result of the operation is the cumulative sum of the multiplication of the corresponding elements in the matrix

3, The quantity product of a matrix

Also known as dot multiplication of matrices , That is, the elements are multiplied

Four , Read and write files

1,save Function is to save an array of data in binary format

Create a new one on your desktop called save Folder .

#np.save(" route \\ Result filename ",arr name )
np.save("C:\\Users\\Yan\\Desktop\\save\\save_arr1",arr)

give the result as follows ：

 

2,savez Function can save multiple arrays to a file

np.savez("C:\\Users\\Yan\\Desktop\\save\\save_arr2",arr1,arr2)

  give the result as follows ：

3,load The function reads data from a binary file

      The extension can be omitted during storage , However, it is not allowed to omit .

np.load("C:\\Users\\Yan\\Desktop\\save\\save_arr1.npy")

  give the result as follows ：

4,savetxt() Function to write an array to a delimited text file .

np.savetxt("C:\\Users\\Yan\\Desktop\\save\\save_arr3.txt",arr,fmt="%d",delimiter="..")

5,loadtxt() Function to load a file into a two-dimensional array .

np.loadtxt("C:\\Users\\Yan\\Desktop\\save\\save_arr3.txt",delimiter="..")
#delimiter Medium is txt How to split storage 

   give the result as follows ：

5、 ... and ,numpy Statistical analysis in

1,sort() Direct sort

2, Indirect sort

argsort()： Returns the subscript of the value sorted in ascending order along the axis

np.argsort(arr1)

lexsort()： Used to sort multiple sequences . Returns the subscript of the sorted value

np.lexsort((arr,arr1))

3, Repetition and de duplication

duplicate removal ：

unique Function to find unique values in an array , And return the results sorted in ascending order

repeat ：

tile() Repeat the entire array

np.tile(arr,2)
# 2 The number of repetitions 

repeat()： Repetitive elements , Repeat by row or column can be specified