Repaso de python.¶

Objetivo: El objetivo de esta sesion es estudiar conceptos basicos de python necesarios para realizar modelos de maching learning

Python Básico
Objetos
Operación filter, map, reduce
Pandas,
Matplotlib , Sea born
Aplicaciones

Python Philosophy..¶

https://www.python.org/dev/peps/pep-0020/#abstract

import this

The Zen of Python, by Tim Peters

Beautiful is better than ugly.
Explicit is better than implicit.
Simple is better than complex.
Complex is better than complicated.
Flat is better than nested.
Sparse is better than dense.
Readability counts.
Special cases aren't special enough to break the rules.
Although practicality beats purity.
Errors should never pass silently.
Unless explicitly silenced.
In the face of ambiguity, refuse the temptation to guess.
There should be one-- and preferably only one --obvious way to do it.
Although that way may not be obvious at first unless you're Dutch.
Now is better than never.
Although never is often better than *right* now.
If the implementation is hard to explain, it's a bad idea.
If the implementation is easy to explain, it may be a good idea.
Namespaces are one honking great idea -- let's do more of those!

Functions, Loops, Conditionals, and list comprehensions¶

my_list = []
for number in range(0, 10):
  if number % 2 == 0:
    my_list.append(number)
my_list

[0, 2, 4, 6, 8]

my_list = [number for number in range(0, 10)  if number %2 ==0]

my_list

[0, 2, 4, 6, 8]

def times_tables():
    """    
    Params:
      --  

    Return:
      -- lst: List

    """
    lst = []
    for i in range(10):
        for j in range (10):
            lst.append(i*j)
    return lst

times_tables() == [j*i for i in range(10) for j in range(10)]

True

Built-in Functions¶

https://docs.python.org/3/library/functions.html

Function map¶

import numpy as np
b = map(lambda x: x**2, range(10))

type(b)

map

list(b)

[0, 1, 4, 9, 16, 25, 36, 49, 64, 81]

sum(b)

def squares_0():
  squares = []
  for x in range(10000):
    squares.append(x**2)
  return squares


# Operation use in big data files
def squares_1():
  return list(map(lambda x: x**2, range(10000)))


def squares_2():
  return [x**2 for x in range(10000)]

squares_0()

import numpy as np
import time

times = [] 
tmax = 10000
for i in range(0,tmax):
  # Common for
  tini0 = time.time()
  squares_0()
  tend0 = time.time()
  # Operation map
  tini1 = time.time()
  squares_1()
  tend1 = time.time()
  # Comprhension expresion 
  tini2 = time.time()
  squares_2()
  tend2 = time.time()    
  times.append([tend0-tini0, tend1-tini1, tend2-tini2] )

t=np.array(times)

---------------------------------------------------------------------------
KeyboardInterrupt                         Traceback (most recent call last)
Input In [14], in <cell line: 3>()
      3 for i in range(0,tmax):
      4   # Common for
      5   tini0 = time.time()
----> 6   squares_0()
      7   tend0 = time.time()
      8   # Operation map

Input In [11], in squares_0()
      2 squares = []
      3 for x in range(10000):
----> 4   squares.append(x**2)
      5 return squares

KeyboardInterrupt: 

mean = np.mean(t, axis = 0)
print(mean)

[0.00034886 0.00033289 0.00029814]

# Magic command
# https://ipython.readthedocs.io/en/stable/interactive/magics.html

%%timeit -n 100 
squares_0()

1000 loops, best of 5: 3.19 ms per loop

%%timeit -n 1000 
squares_1()

1000 loops, best of 5: 3.03 ms per loop

%%timeit -n 1000  
squares_2()

1000 loops, best of 5: 2.79 ms per loop

Lambda function¶

f = lambda x: (x+2) 

f(2)

Filter function¶

Filtros avanzados en python a través de programacion funcional

Ref = https://docs.hektorprofe.net/python/funcionalidades-avanzadas/funcion-filter/,

https://docs.python.org/3/library/functions.html#filter

def multiple(numero):    # Primero declaramos una función condicional
    if numero % 5 == 0:  # Comprobamos si un numero es múltiple de cinco
        return True      # Sólo devolvemos True si lo es

numeros = [2, 5, 10, 23, 50, 33, 5000]
a = filter(multiple, numeros)

<filter at 0x7fe819614110>

list(a)

[5, 10, 50, 5000]

list(a)

[]

FUNCIÓN MAP¶

a = [10.00, 11.00, 12.34, 2.0 ]
b = [9.8, 11.10, 12.34, 2.01 ]

var = map(min, a, b)
var

<map at 0x7fe819614d50>

list(var)

[9.8, 11.0, 12.34, 2.0]

var

<map at 0x7fe819614d50>

list(var)

[]

# Este resultado no muestra nada, ¿porqué?. 
#La variable var ya fue evaluada a través de elementos funcionales
for item in var:
  print(item)

Otro ejemplo con la operación map: Dejar el apellido de las siguientes personas

people = ['Dr. Simon Einstein', 'Dr. Pedro Euler  ', 'Dr. Juan Tesla', 'Dr. Daniel Maxwell']
people[0].split()

['Dr.', 'Simon', 'Einstein']

def split_title_and_name(person):  
    title = person.split()[0]
    lastname = person.split()[-1]
    return f'{title} {lastname}'

last_names = map(split_title_and_name, people)
list(last_names)

['Dr. Einstein', 'Dr. Euler', 'Dr. Tesla', 'Dr. Maxwell']

a = []
for p in people:
  a.append(split_title_and_name(p))
a

['Dr. Einstein', 'Dr. Euler', 'Dr. Tesla', 'Dr. Maxwell']

Tarea 0.1¶

Determinar los primeros 100 numeros impares empleando la funcion map

def impar(x):
  if(x%2!=0):
    return x

q = map(lambda x: 2*x+1 if (2*x+1)<=100 else 0 , range(50))

#list(q)

Objetos¶

class auto:
  """
  Esta clase asigna un color y un tipo 
  a un clase tipo carro
  """
  var = "taller de carros"

  def set_name_tipo(self, new_tipo):
    self.tipo = new_tipo

  def set_name_color(self, new_color ):
    self.color = new_color

carro = auto()
carro.set_name_color="rojo"
carro.set_name_tipo="bus "
print(f"El carro es {carro.set_name_color} y es un  {carro.set_name_tipo} " )

El carro es rojo y es un  bus

class circulo(object):
  def __init__(self, R, posx, posy ):
    self.R1 = R
    self.posx = posx
    self.posy= posy
  
  def Area(self):
    A = np.pi*(self.R1)**2
    return A  
  def perimetro(self):
    return 2*np.pi*self.R1


class circulo_(object):
  def __init__(self):
    self.R1 = None
    self.posx = None
    self.posy= None
  
  def Area(self):
    A = np.pi*(self.R1)**2
    return A
  
  def perimetro(self):
    return 2*np.pi*self.R1
   
   

c = circulo(1, 0, 0)

c.Area()
c.perimetro()

6.283185307179586

cc=circulo_()

cc.posx=1
cc.posy=1
cc.R1=1

cc.R1

cc.perimetro()

6.283185307179586

Tarea 0.2¶

Given a sentence,you task is build a iterator the words Ref = https://www.youtube.com/watch?v=C3Z9lJXI6Qw&ab_channel=CoreySchafer

class Sentence:
    def __init__(self, sentence):
        self.sentence = sentence
        self.index = 0
        self.words = self.sentence.split()

    def __iter__(self):
        return self

    def __next__(self):
        if self.index >= len(self.words):
            raise StopIteration
        index = self.index
        self.index += 1
        return self.words[index]

my_sentence = sentence('This is a test')

print(next(my_sentence))
print(next(my_sentence))
print(next(my_sentence))
print(next(my_sentence))
#print(next(my_sentence))

This
is
a
test

a = Sentence("hola mundo esta es una prueba")

a.__next__()

'hola'

Diccionarios¶

Elementos basicos con diccionarios

students_class = { "Bob": "Physics","Alice": "Physics","Ana": "Biology" }

for i, s in enumerate(students_class):
  print(i, s, students_class[s])

Bob Physics
Alice Physics
Ana Biology

students_class.items()

dict_items([('Bob', 'Physics'), ('Alice', 'Physics'), ('Ana', 'Biology')])

Otra forma de iteraciones para los diccionarios a través del metodo items()

for key, val in students_class.items():
  print(key, val)

Bob Physics
Alice Physics
Ana Biology

Accediendo a los valores del diccionario Accediendo a las claves

metodo keys()
metodo values()

print(students_class.values())
print(students_class.keys())

dict_values(['Physics', 'Physics', 'Biology'])
dict_keys(['Bob', 'Alice', 'Ana'])

Pandas.¶

Series¶

Data Frame¶

import pandas as pd
students_class = { "Bob": "Physics",
                  "Alice": "Chemistry",
                  "Ana": "Biology" }

# Ndarray unidimensional con ejes etiquetados
s = pd.Series(students_class)
s

Bob        Physics
Alice    Chemistry
Ana        Biology
dtype: object

# https://pandas.pydata.org/docs/reference/series.html
print(type(s.index))
s.index

<class 'pandas.core.indexes.base.Index'>

Index(['Bob', 'Alice', 'Ana'], dtype='object')

#Forma de acceder a los elementos con el número del indice
s.iloc[2]

'Biology'

#Forma de acceder a los indices
s.loc["Alice"]

'Chemistry'

s.Bob

'Physics'

Definición clave valor con enteros como clave.¶

class_code = {99:"Physics", 
              100:"Chemistry", 
              101:"English" }

s = pd.Series(class_code)

     Physics
  Chemistry
    English
dtype: object

s.iloc[2]

'English'

s.loc[99]

'Physics'

Tambien podemos definir el objeto Serie a partir de una lista

grades = pd.Series([8,7,10,1])

grades

   8
   7
  10
   1
dtype: int64

for i, g in enumerate(grades):
  print(i,g)

grades.mean()

6.5

grades.describe()

count     4.000000
mean      6.500000
std       3.872983
min       1.000000
25%       5.500000
50%       7.500000
75%       8.500000
max      10.000000
dtype: float64

Definicion a través de un numpy array

x = np.random.randint(0,20, 100)
random_s = pd.Series(x)

random_s

   10
    0
    3
   19
   18
      ..
  19
  12
   7
   2
  11
Length: 100, dtype: int64

random_s.head()

  10
   0
   3
  19
  18
dtype: int64

#Recorrido por las claves y valores, el metodo head es considerado para mostrar pocos valores

for index, values in random_s.head().iteritems():
  print(index, values)

%%timeit -n 100
x = np.random.randint(0,20, 100)
random_s = pd.Series(x)
random_s+=2 # OPeraciones vectoriales a todo el data frame, más eficiente.

#Comparar cuando se tiene un ciclo para realizar la suma, ¿cuál es mas eficiente?

100 loops, best of 5: 163 µs per loop

Agregando nuevos valores con indices diferentes

s = pd.Series([1,2,3,4,9])

s.loc["nuevo"]=2

      1
      2
      3
      4
      9
nuevo    2
dtype: int64

s.loc["nuevo"]

s["nuevo"]

s.iloc[-1]

Otra forma de definir una serie es a través de :

juan_class = pd.Series(["a", "b","c"], index=["0","1","2"])

juan_class

  a
  b
  c
dtype: object

Data Frame¶

Un DataFrame es una lista de series

d1 = { "Name":"Juan", "Topic":"Quantum Mechanics", "Score" : 10}
d2 = { "Name":"Pedro", "Topic":"statistical", "Score" : 10}
d3 = { "Name":"Ana", "Topic":"Clasical Mechanics", "Score" : 10}

record1 = pd.Series(d1)
record2 = pd.Series(d2)
record3 = pd.Series(d3)

# indices con números enteros
df1 = pd.DataFrame( [record1, record2, record3] )
df1

	Name	Topic	Score
0	Juan	Quantum Mechanics	10
1	Pedro	statistical	10
2	Ana	Clasical Mechanics	10

# Asignando nombre a los indices
df2 = pd.DataFrame( [record1, record2, record3] , index = ["UdeA","Unal", "ITM"] )
df2

	Name	Topic	Score
UdeA	Juan	Quantum Mechanics	10
Unal	Pedro	statistical	10
ITM	Ana	Clasical Mechanics	10

#Accediendo a los indices por el nombre
df2.loc["UdeA"]

Name                  Juan
Topic    Quantum Mechanics
Score                   10
Name: UdeA, dtype: object

#Accediendo a los indices por el numero
df2.iloc[0]

Name                  Juan
Topic    Quantum Mechanics
Score                   10
Name: UdeA, dtype: object

#Accediendo a un elemento en particular
df2.loc["UdeA", "Name"]

'Juan'

#Accediendo a algunas columnas del data frame
df2.loc[:, ["Name", "Topic"]]

	Name	Topic
UdeA	Juan	Quantum Mechanics
Unal	Pedro	statistical
ITM	Ana	Clasical Mechanics

Se recomienda crear copias del data frame cuando se esta trabajando con pandas a traves del metodo copy() y no con el operador =, dado que se comparte el mismo espacio de memoria

df2

	Name	Topic	Score
UdeA	Juan	Quantum Mechanics	10
Unal	Pedro	statistical	10
ITM	Ana	Clasical Mechanics	10

a = df2

	Name	Topic	Score
UdeA	Juan	Quantum Mechanics	10
Unal	Pedro	statistical	10
ITM	Ana	Clasical Mechanics	10

a.loc["UdeA", "Name"] = "JuanB"

	Name	Topic	Score
UdeA	JuanB	Quantum Mechanics	10
Unal	Pedro	statistical	10
ITM	Ana	Clasical Mechanics	10

df2

	Name	Topic	Score
UdeA	JuanB	Quantum Mechanics	10
Unal	Pedro	statistical	10
ITM	Ana	Clasical Mechanics	10

b = df2.copy()

Eliminacion de columnas

del b["Topic"]

	Name	Score
UdeA	JuanB	10
Unal	Pedro	10
ITM	Ana	10

Agregando nuevas columnas al data frame

b["Nueva"] = [10, 8, 3]

	Name	Score	Nueva
UdeA	JuanB	10	10
Unal	Pedro	10	8
ITM	Ana	10	3

Tarea 0.3¶

Empleando los siguientes tiempos:

t = np.linspace(0, 2, 1000)

Crear un data frame de pandas para la posicion $y=ho-0.5gt^2$, $g=9.8m/s$, $h = 100 m$
Adicione una nueva columna para la velocidad y la aceleración.

Construya un nuevo data frame desde el tiempo t=0.5s a 1.5s, solo con las posición como funcion del tiempo.

# Exercise
#https://github.com/ajcr/100-pandas-puzzles/blob/master/100-pandas-puzzles.ipynb

Aprendizaje Estatadistico

Repaso de python.

Contents

Repaso de python.¶

Python Philosophy..¶

Functions, Loops, Conditionals, and list comprehensions¶

Built-in Functions¶

Function map¶

Lambda function¶

Filter function¶

FUNCIÓN MAP¶

Tarea 0.1¶

Objetos¶

Tarea 0.2¶

Diccionarios¶

Pandas.¶

Series¶

Data Frame¶

Definición clave valor con enteros como clave.¶

Data Frame¶

Tarea 0.3¶