np.array 코드 요약

Numpy array프로그래밍과 연산에 사용되는 기초 코드들을 하나의 예제 코드로 정리했습니다. 도움이 되시기를 바래요.

#This code shows basic np.array usage with several basic functions
#Print() is used for print result.
#Written by Princess Liatris in 2017-01-11

import numpy as np # import numpy lib as np to use np.array

#01. random nparray declaration
data = 2 + np.random.randn(5,5) + 0 # 5x5 random array declaration. np array need initialization/ sigma * np.random.randn(...) + mu(avg)
print(data) # print random variables

#02. nparray numeric operations
data2 = data * 10
print('')
print(data2) # print random variables
data3 =  data + data
print('')
print(data3) # print random variables

#03. show array shape
print('')
print(data.shape) # shows 5x5 (5,5)
print('')
print(data.dtype) # shows numpy data types(result is float64)

#04. generate ndarray & array copy
data4 = [6,7.5,8,0,1]
arr1 = np.array(data4) #generate new np.array with array function
#print copied dataset
print('')
print(arr1)

#05. list array to multi - dimention array
data5 = [[1,2,3,4],[5,6,7,8]] # generate list array
print('')
print(data5) # print list array

arr2 = np.array(data5)#copy and change as multi-dimention array
print('')
print(arr2) #print changed result with arr2
print('')
print(arr2.ndim) # print dimention with arr2
print('')
print(arr2.shape) # result is (2,4)


#06. np.array initialization
print('')
zeros = np.zeros((2,5))#need 2 (()), filled with 0
print(zeros) #generate 2x5 array with 0
print('')
zeros = np.ones((2,5)) # filled with 1
print(zeros) #generate 2x5 array with 1
print('')
zeros2 = np.empty([2,4])#  filled with non initialized, can give([size], dtype=,order)
print(zeros2)#generate 2x4 array with 'non initialized' data, not empty

#print arranged data
print(np.arange(20))


#07 np.array data types
arr3 = np.array([1,2,3],dtype = np.float64) #numpy's basic dtype is float64
arr4 = np.array([1,2,3],dtype = np.int32)
print('')
print(arr3.dtype) #float64
print(arr4.dtype) #int32

#08. change np.array data type
arr5 = np.array([3.7,1.2,2.6,0.5,12.9])
print('')
print(arr5)

print('')
print(arr5.astype(np.int32))#change data type as int32
arr5 = arr5.astype(np.int32)
arr5 = arr5.astype(np.float64 )#change data type as float64(roleback)
print('')
print(arr5)#changed datatype, loses are not recovered

int_array = np.arange(10)
calibers = np.array([.22,.270,.357,.380,.44,.50],dtype =np.float64)
int_array = int_array.astype(calibers.dtype)#get another array datatype ans assign target array with dtype)
print('')
print(int_array)#changed datatype, float64 type


#09. calc with array and skalar
arr10 = np.array([[1,2,3],[4,5,6]])
print('')
print(arr10)
print('')
print(arr10*arr10)
print('')
print(arr10-arr10)
print('')
print(1/arr10)#broadcasting, skalar values are affect to all variables
print('')
print(arr10 ** 0.5)

#10. slicing ang index
arr10 = np.arange(10)
print(arr10[5])#print 6th element of arr10
print(arr10[5:8])#print 6 - 8th element of arr10

arr10[5:8]=12
print(arr10)#6 - 8th changed to 12
arr_slice = np.array([[1,2,3],[4,5,6],[7,8,9]])
print(arr_slice[2]) # print 1th dimention array with line 3
print(arr_slice[2][2])#print 1 element with 2,2(last element)

arr_slice[2] = 0
print(arr_slice)#all line 2 were changed to 0

print(arr_slice[:2])# :number is remove with : over
print(arr_slice[2:])# number: is eremove with : under

print(arr_slice[:,2:])# : means seelct all, reduce dimentionality

#11. boolean indexing

names = np.array(['bob','joe','will','bob','will','joe','joe'])#generate name index
data = np.random.randn(7,4)# random generate 7 x 4

print(names)
print(data)
print(names == 'bob')# find bob in names
print(data[names =='bob']) # it can use by print other index
print(data[~(names =='bob')]) # choose that not 'bob'
print(data[(names =='bob') | (names =='joe')]) # can use &(and) | (or)


#12. fancy indexing

arr = np.empty((8,4))#generate 8x4

for i in range(8):
    arr[i] = i # imput ith line with i to arr

print(arr)# print arr
print('')
print(arr[[4,3,0,6]])# show 4,3,0,6th row with arr if use - index, reversed

print(arr[[3,3,0,2],[3,3,0,2]])# show (3,3) (3,3) (0,0) (2,2) eleemnts

print(arr[[3,3,0,2]][:,[3,3,0,2]])# show 3,3,0,2 cols and rows, same with arr[np.ix_[3,3,0,2],[3,3,0,2]]

# 13. transpose
arr = np.arange(15).reshape((3,5))# generate 15 list to 3x5 dimention
print(arr)
print(arr.T)#transpose

np.dot(arr.T,arr)# calculate X^tX

print(arr.transpose(1,0))#change lines with 0,1 -> 1,0 line
#print(arr.swapaxes(1,2))# exchange cach rows


#14 universal functions

arr = np.arange(10)
arr2 = np.arange(10) +10
print(np.sqrt(arr))#sqrt
print(np.exp(arr))#exponential

print(np.maximum(arr,arr2))#maximum between 2 arrays

print(np.modf(arr))#used for div result and others


#15 array data processing
points = np.arange(-5,5,0.01)# -5 to 5 with 0.01 increase

xs,ys = np.meshgrid(points, points)#shows with meshgrid form
print(ys)

print(np.where([[True, False], [True, True]],
        [[1, 2], [3, 4]],
       [[9, 8], [7, 6]]))#generate new array from old array

print(arr.mean()) # np mean
print(arr.sum()) # np sum

print(arr.mean(axis = 0)) # np mean wit 1 level low dimentionality


#16 boolean methods
arr = np.random.randn(100)
print((arr > 0).sum()) # number of positive values(print conditional sum)

bools = np.array([False,True,False,False])
#fnd out true
print(bools.any())# there is true
print(bools.all())# there are all true
arr.sort() # sorting array in(), choose dimentionality
print(arr)

#17. np.unique

names = np.array(['bob','bob','angel'])
print(np.unique(names))#find out unique values

values = np.array([6,0,0,3,2,5,6,])

print(np.in1d(values,[2,3,6]))# return true if there is element

# 18 . np.array load and save
arr = np.arange(10)
np.save('some_array',arr) # np.save array with .npy formay
np.savez('array_archive.npz',a = arr, b = arr)#several array saving with compressed npz file format
np.load('some_array.npy')#load .npy format

#np.loadtxt(name, delimiter) - numpy text loader
#np.savetxt(name, delimiter) - numpy text saver

#19. calculus

x = np.array([[1,2],[3,4]])
y = np.array([[1,2],[3,4]])

print(np.dot(x,y))#matrix mul calculation

print(np.random.normal(size =(4,4)))# np randopm with 4x4