Numpy basic learning

Basic operation

a = np.array([1,2,3,4]).reshape((2,2))
b = np.arange(4).reshape((2,2))

c = a-b
d = a*b
#  Multiply the corresponding number of the matrix 
e1 = np.dot(a, b)
e2 = a.dot(b)
#  Matrix multiplication 

print(a<3)
#  You can perform logical operations on matrix elements   Return ture or flose

a = np.random.random((2,4))
# np in random And normal random similar    If random.random((2,4))  production 2 That's ok 4 Column ,0~1 Between random numbers 
#  If the data is already array  Some can be written directly   A.sum()  and np.sum(A)  equally 
a_sum = np.sum(a)
a_min = np.min(a)
a_max = np.max(a)
#  You can just do it   Or column   The biggest and the smallest   Sum up 
a_max_c = np.max(a,axis=1)
a_max_r = np.max(a,axis=0)
#  axis=1  Return the maximum value of each line   Return a list 
#  axis=0  Return the maximum value of each column    Return a list 

print(np.argmax(a))
# a Index value of the maximum value in     also   argmix
print(np.mean(a))
# a The average of 
print(np.average(a))
# average Average 
print(np.median(a))
# median  Median 
print(np.cumsum(a))
# a The accumulation function of   result  [ 2  5  9 14 20 27 35 44 54 65]
print(np.diff(a))
#  Cumulative difference function   The difference between the latter and the former in each line 
print(np.sort(a))
#  and sort Function similar to    Make an arrangement 
print(np.transpose(a))
print(a.T)
#  Both of the above methods are right a To transpose 
print(np.clip(a,5,10))
# clip（ trim ） clip(Array,Array_min,Array_max)  about a Small and medium Array_min Replace it with Array_min,Array_max  For more than replacement 

array Lookup

a = np.arange(3,15).reshape((3,4))
print(a,'\n')
print(a[2])
#  For the first of the matrix indexes [] That's the line ,  The above result is a The second line of 
print(a[1][1])
print(a[1,1])
#  Specific to a certain location   Above results 2 That's ok 2 Number of columns    The above two expressions are the same 
print(a[1,1:4])
#  It's equivalent to slicing  [1:,2]  Second elements 2~5 Column 
print("---------")
for row in a:
    print('\n',row)
# for The loop will print line by line   There is no direct way to print column by column   It can be done to a Transpose it column by column as follows 
print("---------")
for col in a.T:
    print('\n',col)

#  If you want to iterate over all the elements of the matrix , You need to turn it into one line , Use   a.flat（ This is an iterator, not array, If you want to see array want  a.flatten()）
print("---------")
print(a.flatten())
for item in a.flat:
    print(item)

array Assignment

#  assignment  copy
a = [1,2,3,4,5]
b = a
c = b
print(b)
print(c)
#  If you change a The value of or b Value   all abc  Will change   Because if directly  a = b  amount to a And b relation 
a[1] = 10
print(a)
print(b)
#  For those who only want to assign the same value without Association   need  b = a.copy()
q = a.copy()
print('q', q)
a[2] = 9
print(b) # [1, 10, 9, 4, 5]
print(q) # [1, 10, 3, 4, 5]

array Merge

#  Merge the matrix   hstack vstack
a = np.array([1,2,3])
b = np.array([11,22,33])
print(np.vstack((a,b)))  # vertical( vertical ) stack
print(np.hstack((a,b))) # horizontal( level ) stack
print("---------")
#  For a row vector , If you use the previous transpose method （.T or transpose）  Cannot transpose , Because it is not a array
#  Notice that it's ()
print(a.T)   #  Still [1 2 3]
#  Method  np.newaxis()
print("---------")
print(a[:,np.newaxis])
#  Just add it directly after the vector to be transposed , however newaxis() Cannot act on matrix 

#  Multiple array Connect    You can still use vstack perhaps hstack   You can also use concatenate((x,y))  Can pass axis  Choose the direction of the merge 
#  The following two effects are the same 
print(np.vstack((a,b,a)),"\n")
print(np.concatenate((a,b,a))) 

array Segmentation

a = np.arange(12).reshape((3,4))
print(a)
print('\n',np.split(a,2,axis=1))  # If you don't write axis  The default is axis=0（ The matrix is partitioned up and down ）

#  Sometimes I want to divide by different amounts    For example, for the following matrix, you want to divide it vertically into three parts   Sure   arrary_split
# [[ 0  1  2  3]
#  [ 4  5  6  7]
#  [ 8  9 10 11]]
print('\n',np.array_split(a,3,axis=1))
#  But I don't know how to control the desired results , You still need to learn 

#  You can also use vsplit( Vertical segmentation )  and hsplit   It's basically useless , The top is enough 
print(np.hsplit(a,3)) #  be equal to  print(np.split(a,2,axis=1))

Add

# np.where  Is a vectorized version of ternary expressions , and where second , The three parameters need not be arrays , It can be scalar 
a = np.array([1,2,3,4])
b = np.array([6,7,8,9])
c = np.array([True, False, True, False])
result = np.where(c,a,b)  # [1, 7, 3, 9]
#  If you want to replace some numbers in a set of numbers with scalars 
a = np.random.randn(4,4)
a1 = np.where(a > 0, "a", 3)  #  take a Greater than 0 For a Replace less than with 3
a2 = np.where(a < 0, 2, a) #  Only speak a Small and medium 0 Replacement 
# a1 [['3' '3' '3' 'a']
 ['a' '3' 'a' 'a']
 ['a' '3' 'a' 'a']
 ['3' 'a' 'a' 'a']]
# a2 [[2.         2.         2.         0.25344902]
 [2.27838389 2.         1.44099603 1.32540036]
 [0.16570196 2.         1.34380092 2.20212735]
 [2.         1.15632934 2.28351864 1.68879952]]