python_numpy_syntax

gistfile1.txt

import json
import pandas as pd
import numpy as np
import math
from datetime import datetime
from tabulate import tabulate

# Illustration 1. Array initialization: ones/zeros/arrange/linspace
np.arange(0, 10, 2) # array([0, 2, 4, 6, 8])
np.zeros(10)
np.ones(10)
'''
Two rows of zero's:
     array(
        [
            [0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
            [0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]]
       )
'''
np.zeros((2, 10))
np.linspace(0, 100, 11) # array([  0.,  10.,  20.,  30.,  40.,  50.,  60.,  70.,  80.,  90., 100.])
'''
Three rows of random numbers:
    array([[-0.99363132, -1.06340081, -0.63807592,  1.06158869, -0.15740415],
        [ 1.09372848, -2.07485402,  0.70063301,  1.67463863, -0.39733411],
        [ 2.32379695, -0.99970063, -0.55914676, -2.00343542, -0.36605912]])
'''
np.random.seed(48)
np.random.randn(3,5)
np.random.randint(1,100, (2,3)) # Random number between 1-100, in 2 rows, each row 3 elements.


# Illustration 2. Reshape
arr = np.random.randint(0, 100, 10) # Generate 10 random numbers with values between 0-100. Example: array([54, 92, 12, 85, 26, 22,  2, 61, 21,  7])
arr.reshape(2,5) # Split the ten numbers into two rows: Each row five elements.
'''
array([[35, 50, 64, 67,  3],
       [92, 15, 32,  2, 49]])
'''


# Illustration 3. min/max/argmin/argmax - argmin and argmax are index of min and max elements
arr = np.random.randint(0, 50, 10)
print(f"max: {arr.max()}, index: {arr.argmax()}, min: {arr.min()}, index: {arr.argmin()}")


# Illustration 4. Slicing
arr = np.random.randint(0, 100, 10) # List of ten numbers with values between 0-100
arr[7:] # Take from 7th element on


# Illustration 5. Filtering
'''
arr
array([92, 30, 94, 60, 56, 57, 30, 10, 48, 68])

filter - an array of True or False
array([ True, False,  True,  True,  True,  True, False, False, False, True])

arr[filter]
array([92, 94, 60, 56, 57, 68])
'''
arr = np.random.randint(0, 100, 10)
filter = arr>50
arr[filter]


# Illustration 6. Array Operations
arr1 = np.random.randint(0, 100, 10)
arr2 = np.random.randint(0, 100, 10)
arr3 = arr1 + arr2
arr3 = arr3/100 # Instead of looping elements, vectorized operations much faster.
arr3


# Illustration 7. Sum across rows or columns
# array([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24])
arr = np.arange(0, 25) 
'''
array([[ 0,  1,  2,  3,  4],
       [ 5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14],
       [15, 16, 17, 18, 19],
       [20, 21, 22, 23, 24]])
'''
arr = arr.reshape(5,5)
'''
Sum across the rows
    array([50, 55, 60, 65, 70])
'''
arr.sum(axis=0)
'''
Sum across the columns
    array([ 10,  35,  60,  85, 110])
'''
arr.sum(axis=1)


# Illustration 8. statistics: min/max/mean/var/std
arr = np.random.randint(0, 100, 1000) # Generate 1000 numbers, values between 0-100
print(f"max: {arr.max()}, min: {arr.min()}, mean: {arr.mean()}, variance: {arr.var()}, standard deviation: {arr.std()}")
arr


# Illustration 9. Broadcasting
arr1 = np.array([[1, 2, 3]])  # Shape: (1, 3)
arr2 = np.array([[4], [5], [6]])  # Shape: (3, 1)
'''
array([[4+1, 4+2, 4+3],
       [5+1, 5+2, 5+3],
       [6+1, 6+2, 6+3]])

array([[5, 6, 7],
       [6, 7, 8],
       [7, 8, 9]])
'''
arr3 = arr1 + arr2