from datascience import *
import numpy as np

%matplotlib inline
import matplotlib.pyplot as plots
plots.style.use('fivethirtyeight')

Lecture 7¶

Review¶

united_flights = Table.read_table('united.csv')
united_flights.show(2)

Discussion Question: On what fraction of days was flight 73 delayed?

#  (# of times that 73 was delayed) / (# of times that 73 existed)
flight73 = united_flights.where('Flight Number', 73)
flight73.where('Delay', are.above(0)).num_rows / flight73.num_rows

US Census Visualization¶

full = Table.read_table('nc-est2019-agesex-res.csv')
partial = full.where('AGE', are.below(100)).select('SEX', 'AGE', 'POPESTIMATE2019')
us_pop_2019 = partial.relabeled(2, '2019')
us_pop_2019.show(5)

us_pop_2019.where('SEX', 0).plot('AGE', '2019')

us_pop_2019.where('SEX', 0).scatter('AGE', '2019')

What does the expression below compute?

np.average(us_pop_2019.where('SEX', 0).column('AGE'))

Answer:

Not the average age of a person in the US in 2019!
The average of all numbers from 0 to 99
This expression isn’t about the US population at all!

How would you compute the average age of people in the US in 2019?

(Total age of all people) / (# of people)
(Sum over ages of age times # of people of that age) / (# of people)

everyone = us_pop_2019.where('SEX', 0)
total_years = sum(everyone.column('AGE') * everyone.column('2019'))
total_people = sum(everyone.column('2019'))
total_years / total_people

Males vs Females¶

# Let's compare male and female counts per age
males = us_pop_2019.where('SEX', 1).drop('SEX')
females = us_pop_2019.where('SEX', 2).drop('SEX')

pop_2019 = Table().with_columns(
    'Age', males.column('AGE'),
    'Males', males.column('2019'),
    'Females', females.column('2019')
)
pop_2019

pop_2019.plot('Age')

Scatter Plots¶

# Actors and their highest grossing movies
actors = Table.read_table('actors.csv')
actors

actors.scatter('Number of Movies', 'Total Gross')

actors.scatter('Number of Movies', 'Average per Movie')

actors.where('Average per Movie', are.above(400))

Bar Charts¶

# Highest grossing movies as of 2017
top_movies = Table.read_table('top_movies_2017.csv')
top_movies

top10_adjusted = top_movies.take(np.arange(10))
top10_adjusted

# Convert to millions of dollars for readability
millions = np.round(top10_adjusted.column('Gross (Adjusted)') / 1000000, 3)
top10_adjusted = top10_adjusted.with_column('Millions', millions)
top10_adjusted

# A line plot doesn't make sense here: don't do this!
top10_adjusted.plot('Year', 'Millions')

top10_adjusted.barh('Title', 'Millions')

Exercise: Generate the chart shown in the slides: a bar chart of age (# years since release) for the 10 highest grossing movies (non-adjusted).

# Get a table of the top 10 grossing movies (not adjusted for increasing ticket prices)
#  - sort then take *
# Compute the age of each of these movies
#  - get an array of the years they were released and subtract from 2026 *
# Put it all together in a bar chart
#  - use with_column to put ages into the table and then use barh

top10 = top_movies.sort('Gross', descending=True).take(np.arange(10))
ages = 2026 - top10.column('Year')
top10_with_age = top10.with_column('Age in 2026', ages)
top10_with_age.barh('Title', 'Age in 2026')