Python for Finance
Control Flow, data structures and first
application (part 2)
Andras Niedermayer
Outline
1 Control Flow
2 Modules
3 Data types and structures. Working with arrays and matrices.
4 Numpy functions
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 2/34
Functions
• definition with parameters
In [1]: def hello ( name ):
... : print ( ’ Hello ’ + name )
• usage:
In [1]: hello ( ’ Alice ’)
In [2]: hello ( ’ Bob ’)
• even better: document your code with a docstring
In [1]: def hello ( name ):
... : """
... : A function that says hello
... : to ‘ name ‘
... : """
... : print ( ’ Hello ’ + name )
In [2]: help ( hello )
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 3/34
Functions
• return values
def get_answers ( answer_number ):
if answer_number == 1:
return ’ You are the first . ’
elif answer_number == 2:
return ’ Twice as good . ’
else :
return ’ Something else than one ’\
’ or two ’
• usage:
In [1]: print ( get_answer (2))
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 4/34
Functions
• Or pick a random one:
import random
fortune = get_answer ( random . randint (1 ,9))
print ( fortune )
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 5/34
Variable Scope
• one cannot use variables outside of their scope
def spam ():
eggs = 31337
print ( str ( eggs ))
spam ()
print ( eggs )
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 6/34
Variable Scope
• This also true for calls to other functions:
def spam ():
eggs = 31337
bacon ()
print ( eggs )
def bacon ():
ham = 101
eggs = 0
spam ()
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 7/34
Variable Scope
• In constrast to this, one can use global variables in a funtion:
def spam ():
print ( eggs )
eggs = 31337
spam ()
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 8/34
Functions
• The following is bad style, but feasible:
def spam ():
eggs = ’ spam local ’
print ( eggs )
def bacon ():
eggs = ’ bacon local ’
print ( eggs )
spam ()
print ( eggs )
eggs = ’ global ’
bacon ()
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 9/34
Variable Scope
• Usage of the ‘global’ keyword
def spam ():
global eggs
eggs = spam
eggs = ’ global ’
spam ()
print ( eggs )
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 10/34
Our Own Module
• Definition in Python documentation:
“A module is a file containing Python definitions and statements.
The file name is the module name with the suffix .py appended.”
• A module is essentially a file that one imports. The file contains
the functions we can call.
• Here is an example of what we may try to achieve:
In [1]: import sequences
In [2]: # Import the file sequences . py
In [3]: dir ( sequences )
Out [3]: [ ’ __name__ ’ , ’ fib ’ , ’ fib2 ’]
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 11/34
Outline
1 Control Flow
2 Modules
3 Data types and structures. Working with arrays and matrices.
4 Numpy functions
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 12/34
Our Own Module
• Let us create our module “sequences” and save it as
“sequences.py”
""" A module with different sequences """
def fib ( n ):
"""
prints the Fibonacci sequence from 1 to n
"""
a, b = 0, 1
while b < n :
print (b , end = " " )
a, b = b, a+b
print ()
def fib2 ( n ):
"""
returns the Fibonacci sequence until n
"""
res = []
a, b = 0, 1
while b < n :
res . append ( b )
a, b = b, a+b
return res
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 13/34
Our Own Module
• using the module
In [1]: import sequences
In [2]: sequences . __name__
Out [2]: ’ sequences ’
In [3]: sequences . fib (1000)
1 1 2 3 5 8 13 21 34 55 144 233 377 610 987
In [4]: help ( sequences )
In [5]: help ( sequences . fib )
• to automatically reload changes:
In [1]: % load_ext autoreload
In [2]: % autoreload 2
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 14/34
Our Own Module
• making the module executable
• because you want to test the modul
• because you want to run your module without starting the
Python interpreter
• add the following to the end of the sequences.py file:
if __name__ == " __main__ " :
import sys
# Usage : python sequences . py < ENTER >
fib ( int ( sys . argv [1]))
• start the Anaconda prompt, change to the folder in which you
saved sequences.py, e.g. if the folder name is “C:\some\path”,
type
cd ‘‘C:\some\path’’
• then type
python sequences.py 1000
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 15/34
Excercises
14 Create a program that picks a random number between 1 and 2.
The player has to guess the number. The player has 6 attempts
to guess. After each guess, tell the player whether his number
was too high or too low.
15 “Solve” the Syracuse problem numerically: Take an integer
n ≥ 1, repeat the following operation:
• if the number is even then divide it by two
• if the number is odd then multiply it by 3 and add 1
Does the sequence always reach 1?
(“Solve” is in quotation marks, because proving this result in
general is an unsolved problem so far, see
https://en.wikipedia.org/wiki/Collatz_conjecture.
However, we can calculate the solution of the problem for
different numerical examples.)
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 16/34
Excercise 14
• Guessing game guess.py:
"""
Picks a random number and lets
the player guess
"""
# imports
# ask six times
# was the number too high ?
# was the number too low ?
# if neither , exit
# if success ...
# or if failure ...
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 17/34
Excercise 14
• Guessing game guess.py
"""
Picks a random number and lets
the player guess
"""
import random
secret = random . randint (1 ,20)
print ( " I thought of a number between 1 and 20. " )
# Ask the player six times
for number_guess in range (1 ,7):
print ( ’ What number ? ’)
guess = int ( input ())
# number was too low
if guess < secret :
print ( ’ Your guess is too low . ’)
# number was too high
elif guess > secret :
print ( ’ Your guess is too high . ’)
else :
break # otherwise exit
if guess == secret : # if success
print ( ’ Well played . ’)
else : # or if failure ...
print ( ’ No luck ! ’)
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 18/34
Exercise 15
• define following function, run with different values of
starting value and number steps
def syracuse ( starting_value , number_steps ):
u = starting_value
print (u , end = " " )
for n in range ( number_steps ):
if u % 2 == 0:
u = u // 2
else :
u = 3 * u + 1
print (u , end = " " )
if u == 1:
print ( ’ Converged to 1 after {} ’\
’ steps ! ’. format ( n ))
return
print ( ’ Did not converge in {} steps ! ’\
. format ( number_steps ))
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 19/34
Outline
1 Control Flow
2 Modules
3 Data types and structures. Working with arrays and matrices.
4 Numpy functions
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 20/34
Data types in Python
Type Name Example Notes
Integer int a=10 arbitrarily large; a//4=?
Floats float b=0.35 precision issues; b+0.1=?
Strings string c=”it is a string”
• Interesting methods for strings: c.capitalize(), c.split(),
c.replace(a,b)...
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 21/34
Data structures
tuple
Simple collection of arbitrary objects. Limited methods.
t =(1 , 2.5 , " data " ) # t =1 , 2.5 , " data "
Note that indexing starts at zero: t[0]=1.
Two methods:
1 Count: t.count("data")=1
2 Index: t.index(2.5)=1
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 22/34
Data structures
list
A collection of arbitrary objects; many methods available.
l =[1 , 2.5 , " data " ]
Can convert tuples into lists: l=list(t)? Multiple methods:
1 Append (even another list):
l.append([4,3])=[1,2.5,"data",[4,3]]
2 Insert before index:
l.insert(1,’insert’)=[1,’insert’, 2.5,"data",[4,3]]
3 Remove first occurence:
l.remove(2.5)=[1,’insert’,"data",[4,3]]
4 ”Slice”: l[1:3]=[’insert’,"data"]
5 Sort data: l.sort(), or non-mutating version l2=sorted(l)
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 23/34
Data structures
• Excercise: Play hangman
• rules
• The computer chooses a word.
• In each round the player chooses a letter
• If the letter is in the word, it appears.
• If not, then the counter increases and the game appraoches its
end.
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 24/34
Solution
def play_hangman ( word , n ):
guess = len ( word )*[ ’_ ’]
while n >0:
letters = input ( ’ {}. guesses left : ’\
. format ( n ))
letter = letters [0]
if letter in word :
for i in range ( len ( word )):
if word [ i ] == letter :
guess [ i ] = letter
print ( ’ ’. join ( guess ))
if ’_ ’ not in guess :
print ( ’ success ! ’)
return
else :
n -= 1
print ( ’ wrong . {} guesses left . ’\
. format ( n ))
print ( ’ failure ! ’)
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 25/34
Data structures
dict
Dictionaries with key-value stores. Unordered and un-sortable. Maps
(generally) strings into strings or numbers.
d ={ ’ Last ’: ’ Doe ’ , ’ First ’: ’ John ’ , ’ Country ’: ’ UK ’}
Multiple methods:
1 d.keys()=[’Last’, ’First’, ’Country’]
2 d.values()=[’Doe’, ’John’, ’England’]
3 Mapping in a dictionary: d[’Last’]=’Doe’.
4 Setting an item: d[’Country’]=US
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 26/34
Data structures
set
Mathematical sets: unordered collections of objects, repeated only
once.
s1 = set ([ ’a ’ , ’b ’ , ’c ’ , ’d ’ ])
s2 = set ([ ’e ’ , ’b ’ , ’c ’ , ’f ’ ])
Multiple methods:
1 s1.union(s2)={’a’,’b’,’c’,’d’,’e’,’f’}
2 s1.intersection(s2)={’b’,’c’}
3 s1.difference(s2)={’a’,’d’}
4 s1.symmetric difference(s2)={’a’,’d’,’e’,’f’}
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 27/34
Working with matrices: List arrays
v =[0.5 , 0.75 , 1.0 , 1.5 , 2.0] # vector
m =[ v ,v , v ] # a 3 - by -3 matrix
m =[ [0.5 , 0.75 , 1.0 , 1.5 , 2.0]
[0.5 , 0.75 , 1.0 , 1.5 , 2.0]
[0.5 , 0.75 , 1.0 , 1.5 , 2.0]]
1 Easy to select rows or single elements.
Example: m[1] is second row, m[1][0] first element of the second
row.
2 Not easy to select columns! (a ”row” is the primary element of
the list matrix)
3 Works by reference pointers – changes in v are copied
everywhere in m.
Example: v [0] = −2. Try out m =?
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 28/34
Numpy arrays
We will import the numerical Python library: numpy.
import numpy as np
v1 = np . array ([0.5 , 0.75 , 1.0 , 1.5 , 2.0]) # ndarray .
Vector methods for numpy.ndarray:
1 Sum of elements: v1.sum()=5.75.
2 Standard deviation: v1.std()=0.53.
3 Cumulative sum:
v1.cumsum()=array([0.5, 1.25, 2.25, 3.75, 5.75])
4 Scalar multiplication, powers, square root...:
v1*2 = array([1, 1.5, 2.0, 3.0, 4.0])
v1**2 =array([0.25, 0.5625, 1., 2.25, 4.])
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 29/34
Matrix operations
m1 = np . array ([ v1 , v1 *2])
m1 = np . array ([[ 0.5 , 0.75 , 1. , 1.5 , 2.] ,
[ 1. , 1.5 , 2. , 3. , 4.]])
1 Indexing is (row, column): m1[0, 2] is third element of first row.
2 Column sum: m1.sum(axis=0)=array([1.5, 2.25, 3.,
4.5, 6.])
Row sum: m1.sum(axis=1)=?
3 Cumulative sum:
v1.cumsum()=array([0.5, 1.25, 2.25, 3.75, 5.75])
4 Initializing a matrix:
• np.zeros((r,c,z), dtype=’f’, order=’C’) or
np.ones((r,c,z), dtype=’f’, order=’C’).
• Types (optional): i is integer, f is float, b is boolean....
• Order (optional): how to store elements in memory
’C’ is row-wise, ’F’ is column-wise.
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 30/34
Matrix operations (more)
m1 = np . array ([ v1 , v1 *2])
m1 = array ([[0.5 , 0.75 , 1. , 1.5 , 2.] ,
[1. , 1.5 , 2. , 3. , 4.]])
1 Flattening:
m1.ravel()=array([0.5, 0.75, 1., 1.5, 2., 1., 1.5,
2., 3., 4.])
2 Matrix size: m1.shape=(2, 5)
3 Reshape:
m1.reshape(5,-1)=array([[0.5, 0.75], [1., 1.5],
[2., 1.], [1.5, 2.], [3., 4.]])
4 Vertical and horizontal stacking: vstack and hstack.
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 31/34
Vectorization
Advantages of vectorization:
1 compact code, easy to read and understand.
2 faster execution.
v20 = np . array ([2 ,3 ,5])
v21 = np . array ([0.5 ,0.6 ,0.2])
# element - by - element sum
v20 + v21 = array ([2.5 , 3.6 , 5.2])
# broadcasting the scalar .
2* v20 +3= array ([7 , 9 , 13])
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 32/34
Outline
1 Control Flow
2 Modules
3 Data types and structures. Working with arrays and matrices.
4 Numpy functions
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 33/34
Numpy functions
Documentation:
www.docs.scipy.org/doc/numpy/reference/routines.html
Name Description
np.dot(a, b) Dot product of a and b
np.linalg .det(a) Determinant of array a
np.linalg .solve(a, b) Solve linear system ax = b
np.linalg .eig (a) Eigenvalues of matrix a
np.sin(x), np.cos(x).. Trigonometric functions
np.exp(x), np.log (x),
np.power (x1, x2), np.sqrt(x) Arithmetic, exponents, logarithms
np.median(a), np.mean(a)
np.std(a), np.corrcoef (a, b) Summary stats of an array
January 2019 Python for Finance - Lecture 3
Andras Niedermayer - Université Paris-Dauphine 34/34
