Python fundamentals 2

Overview. More core Python. Part 2 of 2.

Python tools. Boolean variables, comparisons, conditionals (if, else), slicing, loops (for), function definitions.

Buzzwords. Code block, data structures, list comprehension, gotcha, PEP8.

Code. Link.

We continue our overview of Python's core language, which lays a foundation for the rest of the course. We go through the material quickly, since we're more interested in the general ideas than the details. You will feel like you're drinking from a fire hose, but it will sink in if you stick with it.

Reminders

Some things from previous chapters that we'll use a lot:

• Assignments and variables. We say we assign what's on the right to the thing on the left: x = 17.4 assigns the number 17.4 to the variable x.

• Strings. Strings are collections of characters in quotes: 'this is a string'.

• Lists. Lists are collections of things in square brackets: [1, 'help', 3.14159].

• Number types: integers vs. floats. Examples of integers include -1, 2, 5, 42. They cannot involve fractions. Floats use decimal points: 12.34. Thus 2 is an integer and 2.0 is a float.

• The print() function. Use print(‘something’, x) to display the value(s) of the object(s) in parentheses.

• The type() function. The command type(x) tells us what kind of object x is. Past examples include integers, floating point numbers, strings, and lists.

• Type conversions. Use str() to convert a float or integer to a string. Use float() or int() to convert a string into a float or integer. Use list() to convert a string to a list of its characters.

• Methods and objects. It's common in Python to work with objects using methods. We apply the method justdoit to the object x by typing x.justdoit.

• Comments. Use the hash symbol # to add comments to your code and explain what you’re doing.

• Tab completion. To find the list of methods available for a hypothetical object x, type x.[tab] in an IPython (Jupyter) notebook. We call that "tab completion."

• Help. We can get help for a function or method foo by typing foo? in the IPython console or foo in the Object explorer. Try each of them with the type() function to remind yourself how this works.

And while we're reviewing: Start Jupyter, open a new file, and save as bootcamp_class_pyfun2.py in your Data_Bootcamp directory/folder.

Dictionaries

The term data structure refers to the organization of a collection of data. Strings and lists are examples. Here we look another one: dictionaries. We won't use them a lot, but when we do they're close to indispensible.

Dictionaries are (unordered) pairs of things defined by curly brackets {}, separated by commas, with the items in each pair separated by colon. For example, a list of first and last names:

names = {'Dave': 'Backus', 'Chase': 'Coleman', 'Spencer': 'Lyon', 'Glenn': 'Okun'}

If we try type(names), the reply is dict, meaning dictionary. The components of each pair are referred to as the key (the first part) and the value (the second). In a real dictionary, the word is the key and the definition is the value. The keys must be unique, but the values need not be.

We access the value from the key with syntax of the form: dict[key]. In the example above, we get Glenn's last name by typing names['Glenn']. (Try it and see.)

We teach ourselves the rest:

Exercise. Print names. Does it come out in the same order we typed it?

Exercise. Construct a dictionary whose keys are the integers 1, 2, and 3 and whose values are the same numbers as words: one, two, three. How would you get the word associated with the key 2?

Exercise. Enter the code

d = {'Donald': 'Duck', 'Mickey': 'Mouse', 'Donald': 'Trump'}
print(d)

What do you see? Why do you think this happened?

Exercise. Describe -- and explain if possible -- the output of these statements:

• list(names)?
• names.keys()?
• list(names.keys())?
• names.values()?

Exercise. Consider the dictionary

data = {'Year': [1990, 2000, 2010], 'GDP':  [8.95, 12.56, 14.78]}

What are the keys here? The values? What do you think this dictionary represents?

Comparisons

Sometimes we want to do one thing if a condition is true, and another if it's false. For example, we might want to use observations for which the date is after January 1980, the country is India, or the population is greater than 5 million -- and not otherwise.

Python does this with comparisons, so called because they involve the comparison of one thing with another. For example, the date of an observation with the date January 1980. The result of a comparison is either True or False. If we assign a comparison to a variable, we refer to it as a Boolean, a name derived from the 18th century mathematician and logician George Boole. This gives us another type to add our collection: float, integer, string, list, dictionary, and now Boolean.

Let's try some simple examples to see what we're dealing with. Suppose we enter 1 > 0 in the IPython console. What does this mean? The input and output look like this:

In [1]:  1 > 0
Out[1]: True

The comparison 1 > 0 is interpreted as a question: Is 1 greater than 0? The answer is True. If we enter 1 < 0 instead,the answer is False.

A comparison is a Python object, but what kind of object is it? We can check with the type() function:

type(1>0)

The answer in this case is bool (that is, Boolean), the name we give to expressions that take the values True and False. (Actually, it says <class 'bool'>, but bool is enough to make the point.)

Python comes with a list of "operators" we can use in comparisons. You can find the complete set in the Python documentation, but common ones include:

• Equals: ==
• Greater than: >
• Greater than or equals: >=
• Does not equal: != (not equals).

We can reverse comparisons (change true to false and vice-versa) with the word not. For example:

In [2]: not 1>0
Out[2]: False

Think about that for a minute and evaluate it piece by piece. 1>0 returns True and not True is false. Additionally, remind yourself that spaces don't matter in Python expressions.

We can do the same thing with variables. Suppose we want to compare the values of variables x and y. Which one is bigger? To see how this works, we run the code

x = 2*3
y = 2**3
print('x greater than y is', x > y)

Here x = 6 and y = 8, so the expression x > y (is x greater than y?) is false.

Exercise. What is 2 >= 1? 2 >= 2? not 2 >= 1? If you're not sure, try them in the IPython console and see what you get.

Exercise. What is 2 + 2 == 4? How about 1 + 3 != 4?

Exercise. What is "Sarah" == 'Sarah'? Can you explain why?

Exercise. What do these comparisons do? Are they true or false? Why?

type('Sarah') == str
type('Sarah') == int
len('Sarah') >= 3

Exercise (challenging). What do you think this code produces?

name1 = 'Chase'
name2 = 'Spencer'
check =  name1 > name2
print(check)

Run it and see if you're right. What type of variable is check? What is its value? Is Chase greater than Spencer?

Conditionals (if and else)

Now that we know how to tell whether a comparison is true or false, we can build that into our code. "Conditional" statements allow us to do different things depending on the result of a comparison or Boolean variable, which we refer to as a condition. The logic looks like this:

• if a condition is true, then do something.
• if a conditions is false, do something else (or do nothing).

To repeat: a condition here is a comparison or Boolean variable and is either true or false.

if statements tell the program what to do if the condition is true:

if 1 > 0:    # read this like "if 1>0 IS TRUE, then do the thing on the next line"
    print('1 is greater than 0')

The syntax here is precise:

• The if statement ends with a colon. That's standard Python syntax, we'll see it again. It's not optional.
• The code that follows is indented exactly four spaces. Also not optional. Jupyter does it automatically.

Both of these features -- a colon at the end of the first line, indent the rest four spaces -- show up in lots of Python code. It's very compact, and the indentation makes the code easy to read.

Exercise. Change the code to

if 1 < 0:
    print('1 is less than 0')

What do you think happens? Try it and see.

Here's another example. Again, we do something if the condition is true, nothing if the condition is false. In this example, the condition is x > 6. If it's true, we print the number. If it's false, we do nothing. The code is

x = 7               # we can change this later and see what happens

if x > 6:
    print('x =', x)

print('Done!')

Here we've set x = 7, which makes the condition x > 6 true. The if statement then directs the program to print x. The blank lines are optional; they make the code easier to read, which is generally a good thing. The statement print('Done!') is just there to tell us that the program finished.

Exercise. What happens if we set x = 4 at the top? How do we know?

else statements tell the program what to do if the condition is false. If we want to do one thing if a condition is true and another if it is false, we would use if for the first and else for the second. The second part has been missing so far. Here's an example:

x = 7

condition = x > 6

if condition:
    print('if branch')             # do if true
    print(condition)
else:
    print('else branch')           # do if false
    print(condition)

The else statement adds the second branch to the decision tree: what to do if the condition is false. Try this with x = 4 and x = 7 to see both branches in action.

Exercise. Take the names name1 and name2, both of them strings. Write a program using if and else that prints the name that comes first in alphabetical order. Test your program with name1 = 'Dave' and name2 = 'Glenn'.

Slicing strings and lists

We can access the elements of strings and lists by specifying the item number in square brackets. This operation is referred to as slicing, probably because we're slicing off pieces, like a cake. The only tricky part of this is remembering that Python starts numbering at zero.

Exercise. Take the string a = 'some'. What is a[1]?

What just happened? Python starts numbering at zero. If we want the first item/letter, we use a[0]. If we want the second, we use a[1]. And so on. We can summarize the numbering convention by writing the word some on a piece of paper. Below it, write the numbers, in order: 0, 1, 2, 3. Label this row "counting forward."

We can also count backward, but again Python has its own numbering convention. If we want the last letter, we use a[-1]. And if we want the one before the last one, we type a[-2]. In this case we get the same answer if we type a[2]. Both give us 'm'.

Let's track this "backward" numbering system in our example. Below the "counting forward" numbers, start another row. Below the letter e write -1. As we move to the left, we type, -2, -3, -4. Label this row "counting backward."

Exercise. Take the string firstname = 'Monty' and write below it the forward and backward counting conventions. What integer would you put in square brackets to extract the third letter (n) under each system?

Exercise. Find the last letter of the string lastname = 'Python'. Find the second to last letter using both the forward and backward counting conventions.

We can do the same thing with lists, but the items here are the elements of a list rather than the characters in a string. The counting works the same way. Let's see if we can teach ourselves.

Exercise. Take the list numberlist = [1, 5, -3]. Use slicing to set a variable first equal to the first item. Set another variable last equal to the last item. Set a third variable named middle equal to the middle item.

More slicing

We've seen how to "slice" (extract) an item from a string or list. Here we'll show how to slice a range of items. For example, slice the last five characters from the string c = 'something'.

Recall that in Python we start counting at zero. If we want the first letter in c, we use c[0]. If we want the second, we use c[1].

If we want more than a single letter, we need to specify both the start and the end. Let's try some examples and see what they do:

c = 'something'
print('c[1] is', c[1])
print('c[1:2] is', c[1:2])
print('c[1:3] is', c[1:3])
print('c[1:] is', c[1:])

Let's go through this line by line:

• The first print statement gives us o, the second letter of something. It's element 1 because we start numbering at zero.
• The next one does the same. Why not two letters? Let's try another one and see.
• The following line gives us om, the second and third letters. Why? Perhaps you figured it out. If not, this is the logic: the second number in 1:3, namely 3, is one more than the end. So the range 1:3 gives us the second and third letters. Confusing, for sure, but that's how it works.
• The last line has no second number. By convention it goes all the way to the end. The slice c[1:] goes from the second letter (the first number 1) to the end, giving us omething.

Some practice:

Exercise. Set lastname = 'Python'. Extract the string 'thon'.

Exercise. Set numlist = [1, 7, 4, 3]. Extract the middle two items and assign them to the variable middle. Extract all but the first item and assign them to the variable allbutfirst. Extract all but the last item and assign them to the variable allbutlast.

Exercise. Take the string c = 'something'. What is c[:3] + c[3:]?

Loops over lists and strings (for)

There are lots of times we want to do the same thing many times, either on one object or on many similar objects. An example of the latter is to print out a list of names, one at a time. An example of the former is to find an answer to progressively higher degrees of accuracy. We repeat an operation as many times as we need to get a desired degree of accuracy. Both situations come up a lot.

Here's an example in which we print all the items in a list, one at a time:

namelist = ['Chase', 'Dave', 'Sarah', 'Spencer']    # creates the list "namelist"
# below, the word "item" is arbitrary. End the line with a colon.
for item in namelist:    # goes through the items in the list one at a time
   print(item)    # indent this line exactly 4 spaces
# if there is code after this, we'd typically leave a blank line in-between

This produces the output

Chase
Dave
Sarah
Spencer

Note that item changes value as we go through the loop. It's a variable whose value actually varies.

We say here that we iterate over the items in the list and refer to the list as an iterable: that is, something we can iterate over. The terminology isn't important, but that's what it means if you run across it.

Exercise. What happens if we replace item with banana in the code above?

Example. We use a loop to compute the sum of the elements of a list of numbers:

numlist = [4, -2, 5]
total = 0
for num in numlist:
    total = total + num

print(total)

The answer (of course) is 7.

Exercise. Adapt the example to compute the average of the elements of numlist.

We can also run loops over the characters in a string. This one prints the letters in a word on separate lines:

word   = 'anything'
for letter in word:
    print(letter)

(You might think we could come up with a more interesting example than this. Sadly no, but we welcome suggestions.)

Example. Here's one that combines a for loop with an if statement to identify and print the vowels in a word:

vowels = 'aeiouy'
word   = 'anything'
for letter in word:
    if letter in vowels:
        print(letter)

(Adapted from SciPy lecture 1.2.) Describe what each line does as well as the overall result.

Example. What about the consonants? Note the word not below:

vowels = 'aeiouy'
word   = 'anything'
for letter in word:
    if letter not in vowels:
        print(letter)

Exercise. Take the list stuff = ['cat', 3.7, 5, 'dog'].

• Write a program that prints the elements of stuff.
• Write a program that tells us the type of each element of stuff.
• Challenging. Write a program that goes through the elements of stuff and prints only the elements that are strings; that is, the print the elements where function type returns the value str.

Loops over counters (range())

We now know how loops work. Here's another version in which we loop over something a fixed number of times. For example, we might want to sum or average the values of a variable. Or value a bond with a fixed number of coupon payments. Or something.

The new ingredient is the range() function. range(n) gives us all the integers (whole numbers) from 0 to n-1. (If that sounds strange, remind yourself how slicing works.) And range(n1, n2) gives us all the whole numbers from n1 to n2-1. We can use it in lots of ways, but loops are a prime example.

Some examples illustrate how this works:

Example. This is one of the simplest uses of range() in a loop:

for number in range(5):     # the variable "number" can be anything
    print(number)

It prints out the numbers 0, 1, 2, 3, and 4. (Ask yourself: Why doesn't it go to 5?) This is like our earlier loops, but range(5) has replaced a list or string as the "iterable."

Here's a minor variant:

for number in range(2, 5):
    print(number)

It prints out the numbers 2, 3, and 4.

Example. We compute and print the squares of integers up to ten. (Paul Ford comments: "Just the sort of practical, useful program that always appears in programming tutorials to address the needs of people who urgently require a list of squares.") We do that with a for loop and the range() function:

for number in range(1, 11):
    square = number**2
    print('Number and its square:', number, square)

Here we have number start at one and work its way up to ten.

Example. Here we compute the sum of integers from one to ten:

total = 0
for num in range(1, 11):
    total = total + num

print(total)

The answer is 55.

Example. Here's one that combines a loop and an if statement:

for num in range(10):
    if num > 5:
       print num

Exercise. Write a loop that computes the first five powers of two.

Example. Consider a bond that pays annual coupons for a given number of years (the maturity) and a principal of 100 at the end. The yield-to-maturity is the rate at which these payments are discounted. Given values for the coupon and the yield, the price of the bond is

maturity = 10
coupon = 5
ytm = 0.05                 # yield to maturity

price = 0
for year in range(1, maturity+1):
    price = price + coupon/((1+ytm)**year)

price = price + 100/((1+ytm)**maturity)
print('The price of the bond is', price)

The answer is 100, which we might know because the coupon and yield are the same once we convert the latter to a percentage. Python gives us 99.99999999999997, which is the computer's version of 100.

Digression. When we originally wrote this code, we used the variable name yield instead of ytm. This was problematic for the following reason: evidently the name yield is reserved for something else. As general rule, it's a good idea to pay attention to issues of this nature, e.g. don't name a variable the same as a built-in function/method, etc.

Loop with condition. Sometimes we want to go through a loop until some condition is met. This combination of a loop and a condition requires an extra level of indenting. It also introduces a new ingredient: the break statement, which tells Python to exit the loop.

Example. Suppose we want to compute the sum of integers until the sum reaches 100. We could use the code

maxnum = 20             # guess of number above our limit

total = 0
for num in range(maxnum):
    total = total + num
    if total > 100:
        break              # exit loop

print('At num =', num, 'we had total =', total)

The if statement starts with a colon and the statement following it (break) is indented four spaces more (eight in total). break is a special command that ends a loop early.

Let's review:

Exercise. In Portugal and Greece, policymakers have suggested reducing their debt by cutting the coupon payments and extending the maturity. How much do we reduce the value of the debt if we reduce the coupons to 2 and increase the maturity to 20?

Exercise. Consider the list namelist = ['Chase', 'Dave', 'Sarah', 'Spencer']. Write a loop that goes through the list until it reaches a name than begins with the letter S. At that point it prints the names and exits the loop.

List comprehensions

That's a mouthful of jargon, but the idea is that we can create lists (and do related things) using implicit loops that we refer to as list comprehensions. This is incredibly useful and shows up a lot in Python code.

Example. Consider the loop above that prints out the elements of the list namelist one at a time:

namelist = ['Chase', 'Dave', 'Sarah', 'Spencer']
for item in namelist:
    print(item)

A list comprehension gives us more compact syntax for the same thing:

[print(item) for item in namelist]

As with loops, the variable item is a dummy: we can use any name we wish. Replace item with your pet's name to see for yourself.

Example. Take the list fruit = ['apple', 'banana', 'clementine']. Here's a list comprehension that creates a new list of capitalized fruits:

FRUIT = [item.upper() for item in fruit]

Try it and see. And think about the loop version a minute to see what we've avoided.

Example. We can do the same with a condition. This one takes the list fruit and creates a new list that contains only those names with six letters or less:

fruit6 = [item for item in fruit if len(item)<=6]

Exercise. Take the list fruit and create a new list with the first letter capitalized. Hint: What method would you use to capitalize a string?

Exercise. Take the list of growth rates g = [0.02, 0.07, 0.07]. Write a list comprehension that multiplies each element by 100 to turn it into a percentage.

Defining our own functions

It's easy to create our own functions -- experienced programmers do it all the time. A common view is that we should never copy lines of code. If we're copying, we're repeating ourselves. What we should do instead is write a function once and use it twice. More than that, breaking a long program into a small number of functions makes the code easier for others to read, which is always a good thing. As we become more comfortable with Python we'll use functions more and more.

Defining functions. The simplest functions have two components: a name (what we call it) and a list of input arguments. Here's an example:

def hello(firstname):               # define the function
    print('Hello,', firstname)

hello('Chase')                      # use the function

Let's go through this line by line:

• The initial def statement defines the function, names it hello, identifies the input as firstname, and ends with a colon (:).
• The following statement(s) are indented the usual four spaces and specify what the function does. In this case, it prints Hello, followed by whatever firstname happens to be. Python understands that the function ends when the indentation ends.
• The last line "calls" the function with input Chase. Note that the name in the function's definition and its use need not be the same.

Function returns. Our function hello has a name (hello) and an input argument (firstname), but returns no output. Output would create a new value that Python could call later in the code, like when you set x = 2 then used x later on. Here we print something but produce no other output.

In other cases, we might want to send output back to the main program. We do that with a return statement, a third component of a function definition. Here's an example

def squareme(number):
    """
    Takes numerical input and returns its square
    """
    return number**2        # this is what the function sends back

square = squareme(7)        # assign the "return" to variable on left
print('The square is', square)

And here's another one:

def combine(first, last):
    """
    Takes strings 'first' and 'last' and returns new string 'last, first'
    """
    lastfirst = last + ', ' + first
    return lastfirst                    # this is what the function sends back

both = combine('Chase', 'Coleman')      # assign the "return" to both
print(both)

Here we return the string 'Coleman, Chase' and assign it to the variable both. Note, too, the comment in triple quotes at the top of the function. That's standard procedure, we recommend it.

The return is an essential component of many functions. Typically when we read the documentation for a function or method, one of the first things we look for is what it returns.

Exercise. Create and test a function that returns an arbitrary power of 2: the input n (an integer) returns the output 2**n. Use n=2 and n=5 as test cases.

Exercise. Create and test a function nextyear that takes an integer year (say 2015) and returns the following year (2016).

Exercise. Use the Object inspector to get the documentation for the built-in function max. If the input is a list of two or more numbers, what does max() return?

Exercise (challenging). Create and test a function that takes a string year (say, '2015') and returns a string of the next year (say, '2016').

Programming style

Yes, style counts. We're not only trying to get something done, we're also communicating with others who may look at our code and possibly use it. A clear style makes that communication more effective.

With that in mind, here are some guidelines we've found useful:

• Put an overall summary of your program at the top in triple quotes. This should include both the purpose of the program and your name. Your email address is optional.
• Lines should be no longer than 79 characters.
• Skip two lines before and after a function definition.
• Skip lines here and there where you think it makes sense.
• Use comments whenever something isn't immediately obvious (and sometimes even when it is).

You can find more along these lines in the classic "PEP8" and Google's style guide.

Some programmers are religious about this. We'd say simply that we want to make our code readable by others.

There's one other thing we often do. If we find documentation online -- at Stack Overflow, for example -- we put a link to it in the code for future reference.

Review

Exercise. What type is each of these expressions? What length?

• 'abcd'
• [1, 3, 5, 7]
• {1: 'one', 2: 'two'}
• 123
• 123.0
• list('abcd')
• range(3)
• list(range(3))

Exercise. Which of the following are True and which are False?

• 2 >= 1
• 2 >= 2
• 2 != 2
• 'this' == "this"
• 'Chase' < 'Dave'
• 'Chase' < 'Dave' or 'Spencer' < 'Glenn'
• 'Chase' < 'Dave' and 'Spencer' < 'Glenn'

Exercise. Take the object numbers = {1: 'one', 2: 'two'}. What type is it? Extract the keys as a list. Extract the values as a list.

Exercise. Write a program that prints the last letter of each item in the list names = ['Chase', Dave', 'Sarah', 'Spencer']. Bonus (optional): Print the last letter only if it's a vowel.

Exercise. Write a function lastletter that extracts the last letter from a string. Use 'Gianluca' as your test case.

Exercise (challenging). Take the list of bond yields y = [0.01, 0.02, 0.03] for maturities of one, two, and three years.

• What happens if you try to multiply all of them by 100 with 100*y?
• How would you accomplish the same task (multiply all the elements of y by 100) with a loop?
• How would you accomplish the same task (multiply all the elements of y by 100) with a list comprehension?

Exercise (challenging). Start with the lists l1 = [1, 2, 3] and l2 = ['one', 'two', 'three'].

• What does list(zip(l1,l2)) do?
• What does dict(list(zip(l1,l2))) do?
• Create a list that contains only the number names in l2 that have three letters.
• Write a list comprehesion that constructs the list of tuples [(1, 1), (2, 4), (3, 9)].
• Convert the list of tuples into a dictionary.

Resources

See the resources in the previous chapter, especially Codecademy. If you work your way up to Advanced Topics, you'll be in good shape for anything that follows.

Additional resources:

• The official Python Tutorial has a nice introduction to "control flow language" that includes comparisons, conditional statements, and loops.
• CodingBat has a great collection of exercises. Significantly more demanding than ours. Runs online.
• Udacity has a free Introduction to Computer Science course that covers Python from a more technical perspective. Recommended for people who want to understand the structure and logic of the language.
• This is way more about comprehensions than you ever wanted to know, but it's so beautifully done you might want to take a look.

One last one, but only if you're curious about floating point numbers. Ok, that's approximately no one. Try this anyway and think about what's going on (it also gives us an idea that we don't want to check for strict equality for floating point numbers -- Better to check whether they are close enough):

0.1 + 0.2 == 0.3

False? More here.