The use of numpy and the summary of common functions

NumPy

ndarray understands multidimensional arrays

import numpy as np

# Generate random multidimensional data of specified dimension
#Mathematical modeling should not be used
data = np.random.rand(2, 3)
print (data)
print (type(data))
#Type is the display data type; shape display dimension; number of ndim dimensions; other functions of type

[[0.46686682 0.68844304 0.76663872]
 [0.70747721 0.47887587 0.25943412]]
<class 'numpy.ndarray'>

ndim, shape and dtype properties

print ('Dimension number', data.ndim)
print ('Dimensions: ', data.shape)
print ('data type: ', data.dtype)

Number of dimensions 2
 Dimensions: (2, 3)
Data type: float64

Create ndarray

'''1. array Establish'''
# list to ndarray
l = range(10)
data = np.array(l)
print (data)
print (data.shape)
print (data.ndim)

[0 1 2 3 4 5 6 7 8 9]
(10,)
1

# Nested sequence to ndarray
l2 = [range(10), range(10)]
#So it forms an array
data = np.array(l2)
print (data)
print (data.shape)

[[0 1 2 3 4 5 6 7 8 9]
 [0 1 2 3 4 5 6 7 8 9]]
(2, 10)

'''2. zeros;ones;empty Establish'''

# np.zeros
zeros_arr = np.zeros((3, 4))
#Pay attention to tuples. Errors are often reported here
# np.ones
ones_arr = np.ones((2, 3))

# np.empty [not all zero, and some random numbers]
empty_arr = np.empty((3, 3))

# np.empty specifies the data type
empty_int_arr = np.empty((3, 3), int)

print (zeros_arr)
print ('-------------')
print (ones_arr)
print ('-------------')
print (empty_arr)
print ('-------------')
print (empty_int_arr)

[[0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]]
-------------
[[1. 1. 1.]
 [1. 1. 1.]]
-------------
[[0.000e+000 0.000e+000 0.000e+000]
 [0.000e+000 0.000e+000 2.174e-321]
 [0.000e+000 0.000e+000 0.000e+000]]
-------------
[[0 0 0]
 [0 0 0]
 [0 0 0]]

# np.arange()
#Creating a series of consecutive numbers is a function similar to range in python in numpy
print (np.arange(10))

[0 1 2 3 4 5 6 7 8 9]

Operation nddarray

Vectorization

# Vector and vector operation
arr = np.array([[1, 2, 3],
                [4, 5, 6]])

print ("Multiply between elements:")
#Pay attention to the operation between discernible matrices. The vector acid here is the same as the broadcast operation
print (arr * arr)

print ("Matrix addition:")
print (arr + arr)

Multiply between elements:
[[ 1  4  9]
 [16 25 36]]
Matrix addition:
[[ 2  4  6]
 [ 8 10 12]]

# Vector and scalar operations
print (1. / arr)
print (2. * arr)

[[1.         0.5        0.33333333]
 [0.25       0.2        0.16666667]]
[[ 2.  4.  6.]
 [ 8. 10. 12.]]

Index and slice

# One-dimensional array
arr1 = np.arange(10)
print (arr1)

print (arr1[2:5])

[0 1 2 3 4 5 6 7 8 9]
[2 3 4]

# Multidimensional array
arr2 = np.arange(12).reshape(3,4)
#To learn how to define multidimensional arrays, range is to form 12 random numbers, and then reshape is to form dimensions
#For example, 3.4.5 = length 4 width 5 height 3
#And points are application functions
print (arr2)

[[ 0  1  2  3]
 [ 4  5  6  7]
 [ 8  9 10 11]]

print (arr2[1])

print (arr2[0:2, 2:])

print (arr2[:, 1:3])

[4 5 6 7]
[[2 3]
 [6 7]]
[[ 1  2]
 [ 5  6]
 [ 9 10]]

# Conditional index

# Find out the data after 2015 in data_arr
data_arr = np.random.rand(3,3)
print (data_arr)

year_arr = np.array([[2000, 2001, 2000],
                     [2005, 2002, 2009],
                     [2001, 2003, 2010]])

is_year_after_2005 = year_arr >= 2005
#: it will be expanded to an array of the same type
print (is_year_after_2005, is_year_after_2005.dtype)

filtered_arr = data_arr[is_year_after_2005]

filtered_arr = data_arr[year_arr >= 2005]
print (filtered_arr)
#Some statements in the middle can be deleted
#The final result is a one-dimensional array, which is very useful for data filtering

[[0.61482194 0.0249229  0.28525661]
 [0.05121173 0.37672803 0.86259463]
 [0.22648329 0.4581513  0.18620441]]
[[False False False]
 [ True False  True]
 [False False  True]] bool
[0.05121173 0.86259463 0.18620441]

# Multiple conditions &|
filtered_arr = data_arr[(year_arr <= 2005) & (year_arr % 2 == 0)]
print (filtered_arr)

[0.61482194 0.28525661 0.37672803]

Transpose = = = transpose

arr = np.random.rand(2,3)
print (arr)
print (arr.transpose())

[[0.01538974 0.47573964 0.90684253]
 [0.93683601 0.64306611 0.63846634]]
[[0.01538974 0.93683601]
 [0.47573964 0.64306611]
 [0.90684253 0.63846634]]

#Conversion of high dimensional array (conversion dimension will be used in the image)
#Don't understand here!!!
arr3d = np.random.rand(2,3,4)
print (arr3d)
print ('----------------------')
print (arr3d.transpose((1,0,2))) # Transposes and definitions of multidimensional arrays do not

[[[0.18074837 0.64652003 0.80527972 0.67800268]
  [0.95766577 0.2498768  0.00304503 0.7058178 ]
  [0.12523549 0.18796252 0.72463798 0.15352211]]

 [[0.38808013 0.31075033 0.53082474 0.32254431]
  [0.6861262  0.02999367 0.70980993 0.09099878]
  [0.14987301 0.78237398 0.90159408 0.82897071]]]
----------------------
[[[0.18074837 0.64652003 0.80527972 0.67800268]
  [0.38808013 0.31075033 0.53082474 0.32254431]]

 [[0.95766577 0.2498768  0.00304503 0.7058178 ]
  [0.6861262  0.02999367 0.70980993 0.09099878]]

 [[0.12523549 0.18796252 0.72463798 0.15352211]
  [0.14987301 0.78237398 0.90159408 0.82897071]]]

ndarray data type conversion = = = astype

zeros_float_arr = np.zeros((3, 4), dtype=np.float64)
print (zeros_float_arr)
print (zeros_float_arr.dtype)

# astype conversion data type
zeros_int_arr = zeros_float_arr.astype(np.int32)
print (zeros_int_arr)
print (zeros_int_arr.dtype)

[[0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]]
float64
[[0 0 0 0]
 [0 0 0 0]
 [0 0 0 0]]
int32

Reading text file

# loadtxt
filename = './presidential_polls.csv'
data_array = np.loadtxt(filename,      # file name
                        delimiter=',', # Specify the element separator inside
                        dtype=str,     # Specify data type
                        usecols=(0,2,3)) # Specifies the column index number to read
print (data_array, data_array.shape)

[['cycle' 'type' 'matchup']
 ['2016' '"polls-plus"' '"Clinton vs. Trump vs. Johnson"']
 ['2016' '"polls-plus"' '"Clinton vs. Trump vs. Johnson"']
 ...
 ['2016' '"polls-only"' '"Clinton vs. Trump vs. Johnson"']
 ['2016' '"polls-only"' '"Clinton vs. Trump vs. Johnson"']
 ['2016' '"polls-only"' '"Clinton vs. Trump vs. Johnson"']] (10237, 3)

# Loadtext, explicitly specifying the type of data in each column
filename = './presidential_polls.csv'
data_array = np.loadtxt(filename,      # file name
                        delimiter=',', # Separator
                        skiprows=1,
                        dtype={'names':('cycle', 'type', 'matchup'),
                               'formats':('i4', 'S15', 'S50')},     # data type
                        usecols=(0,2,3)) # Specifies the column index number to read

print (data_array, data_array.shape) # The result of reading is a one-dimensional array, each element is a tuple

[(2016, b'"polls-plus"', b'"Clinton vs. Trump vs. Johnson"')
 (2016, b'"polls-plus"', b'"Clinton vs. Trump vs. Johnson"')
 (2016, b'"polls-plus"', b'"Clinton vs. Trump vs. Johnson"') ...
 (2016, b'"polls-only"', b'"Clinton vs. Trump vs. Johnson"')
 (2016, b'"polls-only"', b'"Clinton vs. Trump vs. Johnson"')
 (2016, b'"polls-only"', b'"Clinton vs. Trump vs. Johnson"')] (10236,)

Common functions of np

transpose

import numpy as np

arr = np.random.rand(2,3)
print (arr)
print (arr.transpose())

[[0.78485041 0.88817969 0.34809014]
 [0.32744286 0.97539301 0.94401872]]
[[0.78485041 0.32744286]
 [0.88817969 0.97539301]
 [0.34809014 0.94401872]]

#Conversion of high dimensional array (conversion dimension will be used in the image)
#Don't understand here!!!
arr3d = np.random.rand(2,3,4)
print (arr3d)
print ('----------------------')
print (arr3d.transpose((1,0,2))) # Transposes and definitions of multidimensional arrays do not

[[[0.28492549 0.60197236 0.45582367 0.21992479]
  [0.1747163  0.69201365 0.85460359 0.65311699]
  [0.62189644 0.25217555 0.16347156 0.29831219]]

 [[0.42826733 0.81396165 0.187138   0.560564  ]
  [0.10162186 0.66419751 0.03261665 0.06969256]
  [0.55461652 0.55020586 0.50693591 0.31741807]]]
----------------------
[[[0.28492549 0.60197236 0.45582367 0.21992479]
  [0.42826733 0.81396165 0.187138   0.560564  ]]

 [[0.1747163  0.69201365 0.85460359 0.65311699]
  [0.10162186 0.66419751 0.03261665 0.06969256]]

 [[0.62189644 0.25217555 0.16347156 0.29831219]
  [0.55461652 0.55020586 0.50693591 0.31741807]]]

ceil and floor and rint and isnan

arr = np.random.randn(2,3)

print (arr)
print (np.ceil(arr))
#Up nearest integer
print (np.floor(arr))
#Down nearest integer
print (np.rint(arr))
#Rounding
print (np.isnan(arr))
#Determine whether the element is NaN
#There are other functions on the notes

[[ 0.262106   -1.33680008 -1.08562543]
 [ 0.3990978   0.1410074   0.64278274]]
[[ 1. -1. -1.]
 [ 1.  1.  1.]]
[[ 0. -2. -2.]
 [ 0.  0.  0.]]
[[ 0. -1. -1.]
 [ 0.  0.  1.]]
[[False False False]
 [False False False]]

where

arr = np.random.randn(3,4)
print (arr)

np.where(arr > 0, 1, -1)
#(condition, output satisfied, output not satisfied)

[[ 2.04688394  0.48063737  1.20876913 -0.93412937]
 [-0.43427472 -1.47755481  0.36882256 -0.08943138]
 [-0.2847686   0.96915893  0.32641235  0.28346922]]





array([[ 1,  1,  1, -1],
       [-1, -1,  1, -1],
       [-1,  1,  1,  1]])

sum

arr = np.arange(10).reshape(5,2)
print (arr)

print (np.sum(arr))
print (np.sum(arr, axis=0))
print (np.sum(arr, axis=1))

[[0 1]
 [2 3]
 [4 5]
 [6 7]
 [8 9]]
45
[20 25]
[ 1  5  9 13 17]

all and any

import numpy as np
arr = np.random.randn(2,3)
print (arr)

print (np.any(arr > 0))
#One is right
print (np.all(arr > 0))
#All right
'''
·The purpose is to judge whether or not a group of data===It's kind of Boolean
·This can also be applied to pandas Medium DataFrame in
'''

[[-1.020184   -0.48466272 -0.8496271 ]
 [ 0.88815825 -0.81911857  0.64570539]]
True
False





'\n·The purpose is to judge whether or not a group of data===It's kind of Boolean\n·This can also be applied to pandas Medium DataFrame in\n'

unique

arr = np.array([[1, 2, 1], [2, 3, 4]])
print (arr)
print (np.unique(arr))

[[1 2 1]
 [2 3 4]]
[1 2 3 4]