Learning Python for Numeric and Scientific Applications

On By Rodan SottoIn AI

My purpose in learning Python is mainly for numeric and scientific applications, such as AI deep learning.  So my notes will focus mainly on using Python for these types of applications.

The code below demonstrates a little about Python such as some variables and types; lists; subsetting and slicing lists; adding, removing and copying lists; some of its built-in functions and built-in object methods.  Then it demonstrates how to use the NumPy package specifically the NumPy array, some NumPy statistical methods, and how to generate your own sample data.  You can run this code online and I recommend using repl.it.


# Python variables and types
# say we want to calculate bmi = 703 x weight (lbs) / height (in) ^ 2
weight = 150
height = 70
bmi = 703 * (150 / (70 ** 2)) # ** is the operator to calculate powers
print(bmi) # prints value of bmi on the output console
print(type(weight)) # type is int
print(type(height)) # type is int
print(type(bmi)) # type is float
string = "Hello, World!"
string = 'Hello, World!'
boolean = True
print(type(string)) # type is str
print(type(boolean)) # type is bool
# Python lists
# say we want to list the GTA 2016 population in millions
gta_pop = [2.7, 0.6, 1.4, 1.1, 0.5]
# you can have mixed types in the list
gta_pop = ["toronto", 2.7, "durham", 0.6, "peel", 1.4,
"york", 1.1, "halton", 0.5]
# you can have lists inside the list
gta_pop = [["toronto", 2.7], ["durham", 0.6], ["peel", 1.4],
["york", 1.1], ["halton", 0.5]]
print(gta_pop)
print(type(gta_pop)) # type is list
# subsetting or indexing lists
# a 0 index points to the first element
print(gta_pop[0]) # prints toronto population
print(gta_pop[2]) # prints peel population
# a -1 index points to the last element
# you can index the list backwards starting from the last element
print(gta_pop[-1]) # prints halton population
print(gta_pop[-3]) # prints peel population
# slicing lists – selecting multiple elements in the list
# note that the ending index is exclusive
print(gta_pop[1:3]) # prints durham and peel populations
print(gta_pop[2:]) # prints peel, york, and halton populations
print(gta_pop[:3]) # prints toronto, durham, and peel populations
# adding and removing list elements
del(gta_pop[4]) # removes halton population
print(gta_pop)
gta_pop = gta_pop + [["halton", 0.5]] # adds halton population
#gta_pop.append([["halton", 0.5]]) # or you can use append() method
print(gta_pop)
# copying lists by value (not by reference)
gta_pop_copy = list(gta_pop)
gta_pop_copy = gta_pop[:] # or use slicing
print(gta_pop_copy)
# Python built-in functions
print(max([1, 2, 3, 4, 5])) # prints 5
print(round(1.2345, 2)) # prints 1.23
help(max) # opens up documentation on max() function
# Python built-in object methods
mylist = [1, 2, 2, 3, 4, 4, 4, 5]
print(mylist.count(4)) # prints 3 as there are 3 occurrences of element 4 in the list
print(mylist.index(2)) # prints 1 as the index of the first occurrence of element 2 in the list
mystr = "rodan"
print(mystr.capitalize()) # prints Rodan
# Python packages
# importing NumPy package
import numpy
print(numpy.array([1, 2, 3]))
# import specific function in package
from numpy import array as nparray
print(nparray([4, 5, 6]))
# NumPy array is an alternative to Python list
# it provides calculations over entire arrays
# it is easy and fast and specifically created for data science
# it can only contain single type
# say you have 2 datasets containing the weight and height of 5 people
npa_weight = nparray([150, 165, 134, 210, 110])
npa_height = nparray([70, 68, 65, 72, 61])
npa_bmi = 703 * (npa_weight / (npa_height ** 2))
print(npa_bmi) # prints [ 21.52040816 25.08542388 22.29633136 28.47800926 20.78204784]
print(npa_bmi > 22) # prints [False True True True False]
print(npa_bmi[npa_bmi > 22]) # prints [ 25.08542388 22.29633136 28.47800926]
print(type(npa_weight)) # prints numpy.ndarray
npa_2d = nparray([[150, 165, 134, 210, 110],
[70, 68, 65, 72, 61]])
print(npa_2d)
print(npa_2d.shape) # prints (2, 5) – 2 rows and 5 columns
# NumPy array subsetting
print(npa_2d[0]) # prints [150 165 134 210 110], the first row
print(npa_2d[0][2]) # prints 134, the thrid element of the first row
print(npa_2d[0, 2]) # same as above
# NumPy array slicing
print(npa_2d[:, 1:3]) # prints [[165 134] [ 68 65]]
print(npa_2d[0, :]) # prints [150 165 134 210 110], the entire first row
# NumPy statistics
print(numpy.mean(npa_weight)) # prints 153.8
print(numpy.median(npa_weight)) # prints 150.0
print(numpy.std(npa_weight)) # prints 33.4926857687
print(numpy.mean(npa_height)) # prints 67.2
print(numpy.std(npa_height)) # prints 3.86781592116
# NumpPy generate data
# using numpy.random.normal(), passing mean, std, and # of samples, in this case 10 samples
gd_weight = numpy.round(numpy.random.normal(153.8, 33.50, 10), 2)
print(gd_weight) # prints [ 130.67 147.44 153.69 104.84 137.85 164.96 150.62 153.65 194.35 229.11]
#gd_height = numpy.round(numpy)
gd_height = numpy.round(numpy.random.normal(67.2, 3.87, 10), 2)
print(gd_height) # prints [ 72.44 60.82 71.27 70.61 66.06 67.12 68.33 67.81 61.84 74.19]
# using numpy.column_stack() will combine weight and height of each 10 samples
gd_wxh = numpy.column_stack((gd_weight, gd_height))
print(gd_wxh) # prints
#[[ 130.67 72.44]
# [ 147.44 60.82]
# [ 153.69 71.27]
# [ 104.84 70.61]
# [ 137.85 66.06]
# [ 164.96 67.12]
# [ 150.62 68.33]
# [ 153.65 67.81]
# [ 194.35 61.84]
# [ 229.11 74.19]]

view raw

blog_python_numsciapp.py

hosted with ❤ by GitHub

 

Additional References:

Leave a comment