In ds4ci/dplyrExamples: Examples Using dplyr

This example is inspired by a dplyr challenge from Ira Sharenow. Thanks Ira!

Every hour one of my workers shows up at 15 after the hour to take a BART train home. Every hour two BART trains show up randomly in the interval [0,60). I wish to gather data on wait times. The problem is when both trains arrive in [0,15), so I need data from the next hour.

See Ira's full email at end.

Basically we have a simulation problem. We are going to generate two random arrival times for each hour. The number of hours, N_hours, to include in the simulation will determine the accuracy of the simulation.

Once we have the arrival times set up we just need to get the lag between 15 minutes after the hour and the next BART arrival.

The only tricky bit below is handling the case when a random arrival in an hour is before the worker gets to the platform. In that case, we bump the arival time to the next hour with + ifelse(train1_min < worker_minute, 60, 0).

library(dplyr)
library(tidyr)
library(ggplot2)
library(lubridate)
set.seed(1234)                  ## make reproducable
N_hours <- 10000                ## number of hours for the simulation

worker_minute <- 15             ## workers always arive a fixed minutes after the hour
hours <- seq(as.POSIXct("2001-01-01 00:00"), by = "hour", length.out = N_hours)
Waits <- data.frame(hours) %>% 
  mutate(worker_at = hours + dminutes(worker_minute),
         train1_min = runif(N_hours, 0, 59.9999),
         train1_at = hours + dminutes(train1_min + ifelse(train1_min < worker_minute, 60, 0)),
         train1_wait = as.numeric(difftime(train1_at, worker_at, units = "mins")),
         train2_min = runif(N_hours, 0, 59.9999),
         train2_at = hours + dminutes(train2_min + ifelse(train2_min < worker_minute, 60, 0)),
         train2_wait = as.numeric(difftime(train2_at, worker_at, units = "mins")),
         worker_wait = ifelse(train1_wait <= train2_wait, train1_wait, train2_wait)
  )
glimpse(Waits) 
summary(Waits)

Results

summary(Waits$worker_wait)
ggplot(Waits, aes(worker_wait)) + 
  geom_histogram(binwidth = 1) +
  ggtitle("BART Wait Times")

Check distribution of train arrivals within hour

Train_Minutes <- Waits %>% 
  select(hours, train1_min, train2_min) %>% 
  rename(Train1 = train1_min, Train2 = train2_min) %>% 
  gather(Train, Minutes, -hours)
ggplot(Train_Minutes, aes(Minutes, color = Train)) +
  geom_freqpoly(binwidth = 1) +
  xlim(0, 59) + 
  ggtitle("Distribution of Train Arrivals within Hour")

Ira's email

I've taken a different approach than Ira to illustrate a pure dplyr solution.

Jim,

Thanks for giving the talk.

Example 1. Every hour one of my workers shows up at 15 after the hour to take a BART train home. 
Every hour two BART trains show up randomly in the interval [0,60). I wish to gather data on 
wait times. The problem is when both trains arrive in [0,15), so I need data from the next hour.

There is also a problem with &

Thanks for your help.

Ira
---
> library(dplyr)
> over15 = function(x,y) {if( x > 15 & y > x) {res = x - 15} 
+   else if(x > 15 & y < 15) {res = x - 15}
+   else if (y > 15) {res = y - 15} 
+   else res = 500 # needs to be changed. In there to avoid error messages
+   return(res) }
> over15(20, 35)
[1] 5
> over15(8,37)
[1] 22
> over15(54, 28)
[1] 13
> over15(9,7) # The problem case in the data frame
[1] 500
> # row 3 is a problem
> waits = data.frame(Hour = c(1,2,3,4), Arrival = rep(15,4), Train1 = c(20, 8, 2, 5), 
  Train2 = c(37,20,8,53))
> waits  
  Hour Arrival Train1 Train2
1    1      15     20     37
2    2      15      8     20
3    3      15      2      8
4    4      15      5     53
> waits = tbl_df(waits)
> waits
Source: local data frame [4 x 4]

   Hour Arrival Train1 Train2
  (dbl)   (dbl)  (dbl)  (dbl)
1     1      15     20     37
2     2      15      8     20
3     3      15      2      8
4     4      15      5     53
> mutate(waits,
+        delay = over15(Train1, Train2))
Source: local data frame [4 x 5]

   Hour Arrival Train1 Train2 delay
  (dbl)   (dbl)  (dbl)  (dbl) (dbl)
1     1      15     20     37     5
2     2      15      8     20    -7
3     3      15      2      8   -13
4     4      15      5     53   -10
Warning message:
In if (x > 15 & y > x) { :
  the condition has length > 1 and only the first element will be used
> mapply(over15, waits$Train1, waits$Train2)
[1]   5   5 500  38

ds4ci/dplyrExamples documentation built on May 15, 2019, 2:56 p.m.