Note: We are using Python here, not Python 2. This means we are using Python 3!
Note: This lecture comprises only those parts of Python that are fundamental to Machine Learning purposes.
Note: Stop writing, listen to me.

Why do I have to learn Python?



In [ ]:

    
from IPython.display import Image
Image('images/mem0.jpg')



In [ ]:

    
Image('images/mem1.jpg')



In [ ]:

    
Image('images/C++_machine_learning.png')



In [ ]:

    
Image('images/Java_machine_learning.png')



In [ ]:

    
Image('images/Python_machine_learning.png')



In [ ]:

    
Image('images/R_machine_learning.png')

Python is an interpreted language

fire up an interpreter
assign a value to a variable
print something
switch to ipython

Fundamentals

Python uses whitespaces (tabs and spaces) to structure code rather than braces.
Terminating semicolons are not required.



In [ ]:

    
def minimum(x, y):
    if x < y:
        return x
    else:
        return y

Why should one do this when designing a language?

When putting several statements into one line you must use semicolons to seperate them.



In [ ]:

    
a = 42; b = 7; c = 23



In [ ]:

    
print(a, b)

But for the sake of readability you shouldn't do so.

Comments



In [ ]:

    
# This is a comment

#
# This is a multi line comment
#

Functions and methods

Functions are called using parentheses with zero or more arguments. A value may be returned.



In [ ]:

    
foo = min(2, 3)
foo

Methods are called using this syntax



In [ ]:

    
obj.method(parameters)

Positional and keyword arguments



In [ ]:

    
f(a, b, c=5, d='foo')

is equal to



In [ ]:

    
f(a, b, d='foo', c=5)

References



In [ ]:

    
a = [1, 2, 3]
b = a
b.append(4)



In [ ]:

    
b



In [ ]:

    
a

functions are also called by reference.

Dynamic references

In contrast to Java, Python references are typeless. There is no problem with this:



In [ ]:

    
a = 5
print(type(a))
a = 'foo'
print(type(a))

Strong typing



In [ ]:

    
5 + '5'

Types can be checked using isinstance



In [ ]:

    
isinstance(4, int)



In [ ]:

    
isinstance(4.5, int)

You can also pass it a list of types



In [ ]:

    
isinstance(4.5, (int, float))



In [ ]:

    
minimum.bar = lambda x: x
minimum.bar('foo')

Attributes and methods

Attributes and methods are accessed using the obj.attribute or the obj.methods() notation.



In [ ]:

    
a = 'foo'
a.count('o')



In [ ]:

    
a.<TAB>

Imports

Python modules are simple <name>.py files container functions, classes and variables.



In [1]:

    
%%bash
ls -la material/foo.py









    



-rw-rw-r-- 1 user user 49 Mär 24 20:25 material/foo.py



In [2]:

    
%%bash
cat material/foo.py









    



PI = 3.14159

def f():
    return 'Hello, world'



In [ ]:

    
from foo import PI as pi
from foo import f
import foo



In [ ]:

    
f()



In [ ]:

    
pi



In [ ]:

    
foo.f()



In [ ]:

    
foo.PI

Operators

Most operators work as you might expect



In [ ]:

    
7 + 4



In [ ]:

    
5 / 2



In [ ]:

    
4 < 3



In [ ]:

    
4 ** 2



In [ ]:

    
id('this is a somewhat long text')



In [ ]:

    
id('this is a somewhat long text')

Equality

== tests for equivalence
is tests for identity



In [ ]:

    
10000000 + 10000000 is 20000000



In [ ]:

    
10000000 + 10000000 == 20000000

Mutability of objects

In Python there are mutable objects like lists, dicts and user-defined types (classes).



In [ ]:

    
a_list = [0, 1, 2]
a_list



In [ ]:

    
a_list[1] = 3
a_list

But there are also immutable objects like 'tuples' and strings



In [ ]:

    
a_tuple = (0, 1, 2)
a_tuple



In [ ]:

    
a_tuple[0] = 4



In [ ]:

    
foo = 'bar'



In [ ]:

    
id(foo)



In [ ]:

    
foo = foo + 'bar'



In [ ]:

    
id(foo)

Scalar Types

Python has a couple of build in scalar types:

int (32-bit, with automatic conversion to 64-bit)
bool
None
float (64-bit)
str



In [ ]:

    
4



In [ ]:

    
4 / 2



In [ ]:

    
7.3e-4



In [ ]:

    
float(43)



In [ ]:

    
str('foobar')



In [ ]:

    
'foobar'



In [ ]:

    
"foobar"



In [ ]:

    
""" multiline
foobar
"""



In [ ]:

    
foo = 'bar'
foo[1]



In [ ]:

    
str(4)



In [ ]:

    
int('4')



In [ ]:

    
s = 'foo bar'



In [ ]:

    
s[0:3]



In [ ]:

    
list(s)



In [ ]:

    
s = 'bob\'s bar'
print(s)



In [ ]:

    
a = 'foo'
b = 'bar'
a + b



In [ ]:

    
pi = 3.14159
'pi is {:.2f}'.format(pi)