Chris Albon

Two Way Frequency Table

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

# Create some data
A <- c("yes", "no","yes", "no","yes", "no","yes", "no")
B <- c("male", "female","female", "male","male", "male","male", "male")

# A will be rows, B will be columns
mytable <- table(A,B)

# print table
mytable

     B
A     female male
  no       1    3
  yes      1    3

# A frequencies (summed over B)
margin.table(mytable, 1)

A
 no yes
  4   4

# B frequencies (summed over A)
margin.table(mytable, 2)

B
female   male
     2      6

# cell percentages
prop.table(mytable)

     B
A     female  male
  no   0.125 0.375
  yes  0.125 0.375

# row percentages
prop.table(mytable, 1)

     B
A     female male
  no    0.25 0.75
  yes   0.25 0.75

# column percentages
prop.table(mytable, 2)

     B
A     female male
  no     0.5  0.5
  yes    0.5  0.5

2D Density Plot

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

Original source: r graphics cookbook

# load the gcookbook package for the data
library(gcookbook)

# load the ggplot2 package
library(ggplot2)

# reset the graphing device
dev.off()

# create the ggplot2 data
p <- ggplot(faithful, aes(x = eruptions, y = waiting)) +
  # add a layer with the points
  geom_point() +
  # and a layer for the density heatmap with the alpha and the color determined by density (the .. refers to the fact that density is a variable that was created inside the ggplot() function)
  stat_density2d(aes(alpha=..density.., fill=..density..), geom="tile", contour=FALSE)

null device
          1

Adding Time To A Date

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

# load the lubridate package
library(lubridate)

# create a date variable
date.ex <- dmy("1/1/2001")

date.ex

[1] "2001-01-01 UTC"

# add 45 days to a date
date.ex.2 <- date.ex + days(45)

date.ex.2

[1] "2001-02-15 UTC"

# add six weeks to a date
date.ex.3 <- date.ex + weeks(6)

date.ex.3

[1] "2001-02-12 UTC"

Adding labels to a ggplot2 bar graph

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

original source: r graphics cookbook

# load the ggplot2 package
library(ggplot2)

# load the gcookbook package
library(gcookbook)

Below the top

# create a ggplot data
ggplot(cabbage_exp, aes(x=interaction(Date, Cultivar), y=Weight)) +
  # draw the bar plot
  geom_bar(stat="identity") +
  # create the weight text above the bar in white
  geom_text(aes(label=Weight), vjust=1.5, colour="white")

Above the top

# create a ggplot data
ggplot(cabbage_exp, aes(x=interaction(Date, Cultivar), y=Weight)) +
  # draw the bar plot
  geom_bar(stat="identity") +
  # create the weight text below the bar in white
  geom_text(aes(label=Weight), vjust=-0.2)

Labels on a grouped bar chart

# create the ggplot data for a grouped bar chart
ggplot(cabbage_exp, aes(x=Date, y=Weight, fill=Cultivar)) +
  # plot the bars
  geom_bar(stat="identity", position="dodge") +
  # create the label, "dodged" to fit the bars
  geom_text(aes(label=Weight), vjust=1.5, colour="white",
            position=position_dodge(.9), size=3)

ymax not defined: adjusting position using y instead

Adding Levels To A Factor

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

# create simulated distract name data
district <- c("NORTH", "NORTHWEST", "CENTRAL", "SOUTH", "SOUTHWEST", "EAST")

# remake district categories with the combination of district categories and a new SOUTH CENTRAL category
levels(district) <- c(district, "SOUTH CENTRAL")

levels(district)

[1] "NORTH"         "NORTHWEST"     "CENTRAL"       "SOUTH"        
[5] "SOUTHWEST"     "EAST"          "SOUTH CENTRAL"

Aggregate Data By Week Or Month

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

original source: http://stackoverflow.com/questions/19716244/aggregate-data-by-week-month-etc-in-r?lq=1

# load the xts package
library(xts)

Loading required package: zoo

Attaching package: ‘zoo’

The following objects are masked from ‘package:base’:

    as.Date, as.Date.numeric

Create some simulated data

# create an element for every year between two dates
date <- seq(as.Date("2006-01-01"), as.Date("2007-01-01"), by = "1 day")

# create some simulated values
score <- runif(366)

# create a zoo time series object of score and ata
zoo <- zoo(score, date)

Create some averages

# create a weekly average
weekly.avg <- apply.weekly(zoo, mean)

weekly.avg

2006-01-01 2006-01-08 2006-01-15 2006-01-22 2006-01-29 2006-02-05 2006-02-12
 0.6463105  0.3696941  0.4492466  0.5587588  0.3330893  0.7490642  0.3463500
2006-02-19 2006-02-26 2006-03-05 2006-03-12 2006-03-19 2006-03-26 2006-04-02
 0.4594144  0.3015816  0.5016827  0.3824588  0.4501046  0.5086366  0.6927037
2006-04-09 2006-04-16 2006-04-23 2006-04-30 2006-05-07 2006-05-14 2006-05-21
 0.5238080  0.6618441  0.4366701  0.6187016  0.6110044  0.5724795  0.5267836
2006-05-28 2006-06-04 2006-06-11 2006-06-18 2006-06-25 2006-07-02 2006-07-09
 0.4003268  0.3999404  0.6366840  0.4546525  0.5675619  0.4411083  0.5747285
2006-07-16 2006-07-23 2006-07-30 2006-08-06 2006-08-13 2006-08-20 2006-08-27
 0.4136250  0.4936679  0.4814989  0.4419165  0.3644543  0.6385395  0.5230308
2006-09-03 2006-09-10 2006-09-17 2006-09-24 2006-10-01 2006-10-08 2006-10-15
 0.5259253  0.5474812  0.4658602  0.4771834  0.6106620  0.4471343  0.4576065
2006-10-22 2006-10-29 2006-11-05 2006-11-12 2006-11-19 2006-11-26 2006-12-03
 0.6124155  0.5418694  0.3136825  0.4227544  0.2406943  0.3723846  0.6079556
2006-12-10 2006-12-17 2006-12-24 2006-12-31 2007-01-01
 0.5289365  0.4426345  0.6362102  0.4849858  0.1102631

# create a monthly average
monthly.avg <- apply.monthly(zoo, mean)

monthly.avg

2006-01-31 2006-02-28 2006-03-31 2006-04-30 2006-05-31 2006-06-30 2006-07-31
 0.4636550  0.4345047  0.4883293  0.5791776  0.5045688  0.5301847  0.4698487
2006-08-31 2006-09-30 2006-10-31 2006-11-30 2006-12-31 2007-01-01
 0.5033712  0.5218908  0.5171836  0.3661705  0.5341010  0.1102631

The Probability An Economy Seat Is An Aisle?

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

There are two types of people in the world, aisle seaters and window seaters. I am an aisle seater, nothing is worse than limited bathroom access on a long flight. The first thing I do when I get my ticket is check to see if I have a window seat. If not, I immediately head over to the airline counter and try to get one.

Last flight, on Turkish Airlines, I ran into a curious situation. I recieved my boarding pass with my seat number, 18C, but the ticket did not specify if C was an aisle seat or not. Making matters worse, the airline counter was swamped with a few dozen people. So I asked myself: given only the seat letter, C, what is the probability that it is an aisle seat?

Later, on the flight, I decided to find out.

Preliminaries

# Import required modules
import pandas as pd
import numpy as np

# Set plots to display in the iPython notebook
%matplotlib inline

Setup possible seat configurations

I am a pretty frequently flyer on a variety of airlines and aircraft. There are a variety of seating configurations out there, but typically they follow some basic rules:

No window cluster of seats has more than three seats.
On small slights with three seats, the single seat is on the left side.
No flight has more than nine rows.

Based on these rules, here are the "typical" seating configurations from aircraft with between two and nine seats per row. A '1' codifies that a seat is an aisle seat, a '0' codifies that it is a non-aisle seat (i.e. window or middle), and 'np.nan' denotes that the aircraft has less than nine seats (this is so all the list lengths are the same).

# An aircraft with two seats per row
rows2 = [1,1, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan]

# An aircraft with three seats per row
rows3 = [1,1,0, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan,]

# An aircraft with four seats per row
rows4 = [0,1,1,0, np.nan, np.nan, np.nan, np.nan, np.nan]

# An aircraft with five seats per row
rows5 = [0,1,1,0,0, np.nan, np.nan,np.nan, np.nan]

# An aircraft with six seats per row
rows6 = [0,1,1,0,1,0, np.nan, np.nan, np.nan]

# An aircraft with seven seats per row
rows7 = [0,1,1,0,1,1,0, np.nan, np.nan]

# An aircraft with eight seats per row
rows8 = [0,0,1,1,1,1,0,0, np.nan]

# An aircraft with nine seats per row
rows9 = [0,0,1,1,0,1,1,0,0]

For example, in an aircraft with five seats per row, rows5, the seating arrangement would be:

window
aisle
aisle
middle
window
no seat
no seat
no seat
no seat

Next, I'm take advantage of pandas row summation options, but to do this I need to wrangle the data into a pandas dataframe. Essentially I am using the pandas dataframe as a matrix.

# Create a list variable of all possible aircraft configurations
seating_map = [rows2, rows3, rows4, rows5, rows6, rows7, rows8, rows9]

# Create a dataframe from the seating_map variable
df = pd.DataFrame(seating_map,
                  columns=['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I'],
                  index=['rows2', 'rows3', 'rows4', 'rows5', 'rows6', 'rows7', 'rows8', 'rows9'])

Here is all the data we need to construct our probabilities. The columns represent individual seat letters (A, B, etc.) while the rows represent the number of seats-per-row in the aircraft.

# View the dataframe
df

	A	B	C	D	E	F	G	H	I
rows2	1	1	NaN	NaN	NaN	NaN	NaN	NaN	NaN
rows3	1	1	0	NaN	NaN	NaN	NaN	NaN	NaN
rows4	0	1	1	0	NaN	NaN	NaN	NaN	NaN
rows5	0	1	1	0	0	NaN	NaN	NaN	NaN
rows6	0	1	1	0	1	0	NaN	NaN	NaN
rows7	0	1	1	0	1	1	0	NaN	NaN
rows8	0	0	1	1	1	1	0	0	NaN
rows9	0	0	1	1	0	1	1	0	0

Calculate aisle probability

Because each aircraft seats-per-row configuration (i.e. row) is binary (1 if aisle, 0 if non-aisle), the probability that a seat is an aisle is simply the mean value of each seat letter (i.e. column).

# Create a list wherein each element is the mean value of a column
aisle_probability = [df['A'].mean(),
                     df['B'].mean(),
                     df['C'].mean(),
                     df['D'].mean(),
                     df['E'].mean(),
                     df['F'].mean(),
                     df['G'].mean(),
                     df['H'].mean(),
                     df['I'].mean()]

# Display the variable
aisle_probability

[0.25,
 0.75,
 0.8571428571428571,
 0.33333333333333331,
 0.59999999999999998,
 0.75,
 0.33333333333333331,
 0.0,
 0.0]

So there you have it, the probability that each seat letter is an aisle. However, we can make the presentation a little more intituative.

Visualize seat letter probabilities

The most obvious visualization to convey the probabilities would be seat letters on the x-axis and probabilities on the y-axis. Panda's plot function makes that easy.

# Create a list of strings to use as the x-axis labels
seats = ['Seat A', 'Seat B', 'Seat C', 'Seat D',
         'Seat E', 'Seat F', 'Seat G', 'Seat H', 'Seat I']

# Plot the probabilities, using 'seats' as the index as a bar chart
pd.Series(aisle_probability, index=seats).plot(kind='bar', # set y to range between 0 and 1
                                                    ylim=[0,1],
                                                    # set the figure size
                                                    figsize=[10,6],
                                                    # set the figure title
                                                    title='Probabilty of being an Aisle Seat in Economy Class')

<matplotlib.axes._subplots.AxesSubplot at 0x1078300f0>

So there we have it! If given a boarding pass with seat C you have a 86% probability of being in an aisle seat!

I hope this was helpful!

All Combinations For A List Of Objects

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

Preliminary

# Import combinations with replacements from itertools
from itertools import combinations_with_replacement

Create a list of objects

# Create a list of objects to combine
list_of_objects = ['warplanes', 'armor', 'infantry']

Find all combinations (with replacement) for the list

# Create an empty list object to hold the results of the loop
combinations = []

# Create a loop for every item in the length of list_of_objects, that,
for i in list(range(len(list_of_objects))):
    # Finds every combination (with replacement) for each object in the list
    combinations.append(list(combinations_with_replacement(list_of_objects, i+1)))

# View the results
combinations

[[('warplanes',), ('armor',), ('infantry',)],
 [('warplanes', 'warplanes'),
  ('warplanes', 'armor'),
  ('warplanes', 'infantry'),
  ('armor', 'armor'),
  ('armor', 'infantry'),
  ('infantry', 'infantry')],
 [('warplanes', 'warplanes', 'warplanes'),
  ('warplanes', 'warplanes', 'armor'),
  ('warplanes', 'warplanes', 'infantry'),
  ('warplanes', 'armor', 'armor'),
  ('warplanes', 'armor', 'infantry'),
  ('warplanes', 'infantry', 'infantry'),
  ('armor', 'armor', 'armor'),
  ('armor', 'armor', 'infantry'),
  ('armor', 'infantry', 'infantry'),
  ('infantry', 'infantry', 'infantry')]]

# Flatten the list of lists into just a list
combinations = [i for row in combinations for i in row]

# View the results
combinations

[('warplanes',),
 ('armor',),
 ('infantry',),
 ('warplanes', 'warplanes'),
 ('warplanes', 'armor'),
 ('warplanes', 'infantry'),
 ('armor', 'armor'),
 ('armor', 'infantry'),
 ('infantry', 'infantry'),
 ('warplanes', 'warplanes', 'warplanes'),
 ('warplanes', 'warplanes', 'armor'),
 ('warplanes', 'warplanes', 'infantry'),
 ('warplanes', 'armor', 'armor'),
 ('warplanes', 'armor', 'infantry'),
 ('warplanes', 'infantry', 'infantry'),
 ('armor', 'armor', 'armor'),
 ('armor', 'armor', 'infantry'),
 ('armor', 'infantry', 'infantry'),
 ('infantry', 'infantry', 'infantry')]

Annotating Plots

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

Original source: r graphics cookbook

# load the gcookbook package for the data
library(gcookbook)

# load the ggplot2 package
library(ggplot2)

# reset the graphing device
dev.off()

quartz_off_screen
                3

# create the ggplot2 data
p <- ggplot(faithful, aes(x = eruptions, y = waiting))

Add Text

# create the ggplot2 plot
p + geom_point() +
  # add text
  annotate("text", x = 3, y = 48, label="Group 1", family="serif", fontface="italic", colour="darkred", size=6) +
  # add more text
  annotate("text", x = 4.5, y = 66, label="Group 2", family="serif", fontface="italic", colour="darkred", size=6)

# Add Mathematical Expressions

# create the ggplot2 plot
p + geom_point() +
  # add the formula, parse=TRUE turns the next into a formula
  annotate("text", x = 4.5, y = 66, parse = TRUE, label = "frac(1, sqrt(2 * pi)) * e ^ {-x^2 / 2}")

Add Mathematical Expressions

# create the ggplot2 plot
p + geom_point() +
  # add the formula, parse=TRUE turns the next into a formula
  annotate("text", x = 4.5, y = 66, parse = TRUE, label = "frac(1, sqrt(2 * pi)) * e ^ {-x^2 / 2}")

Add Lines

# load the grid package to create the flat ends of the line seqment and arrow
library(grid)

# create the ggplot2 plot
p + geom_point() +
  # add a horizontal line at y = 66
  geom_hline(yintercept = 66) +
  # add a vertical line at 3 = 3
  geom_vline(xintercept = 3) +
  # add an angled line
  geom_abline(intercept = 37.4, slope = 9) +
  # add a line segment
  annotate("segment", x = 1, xend = 2.5, y = 75, yend = 75, arrow=arrow(ends="both", angle=90, length=unit(.2,"cm"))) +
  # add an arrow
  annotate("segment", x = 4, xend = 5, y = 60, yend = 55, colour="blue", size=2, arrow=arrow())

Add A Shaded Rectangle

# create the ggplot2 plot
p + geom_point() +
  # add a shaped blue rectangle
  annotate("rect", xmin=1, xmax=3, ymin=40, ymax=100, alpha=.1, fill="blue")

Apply A Function On Every Row Of A Dataframe

Chris Albon — Sun, 01 May 2016 12:00:00 -0700

original source: http://stackoverflow.com/questions/2074606/doing-a-plyr-operation-on-every-row-of-a-data-frame-in-r

# Load packages
library(plyr)

# create a simulated dataframe
x <- rnorm(10)
y <- rnorm(10)
df <- data.frame(x,y)

# array to dataframe apply, on df, for each row, apply transform() to create a variable called "max" whose values are the maximum value of x or y (whichever is higher).
adply(df, 1, transform, max = max(x, y))

            x          y        max
1  -1.0286311  0.6621974  0.6621974
2   0.5466022  0.2977963  0.5466022
3  -0.6559125 -2.0830247 -0.6559125
4  -1.6942847 -0.2205220 -0.2205220
5   2.2678281 -0.4791234  2.2678281
6  -1.6849528 -0.4873940 -0.4873940
7   1.1627351  0.5137251  1.1627351
8   1.4182618  0.9697840  1.4182618
9   0.2025052 -0.3519337  0.2025052
10 -0.7100003 -0.6827529 -0.6827529