Nick Boonstra Challenge 2

Introduction

For today’s challenge, I will be reading in and wrangling data from the “hotel_bookings” dataset.

Setup Chunk

First, my R setup chunk:

Code

library(tidyverse)

knitr::opts_chunk$set(echo = TRUE, warning=FALSE, message=FALSE)

Reading in the data

Reading in the data was a fairly straightforward process:

Code

hotels<-read_csv("_data/hotel_bookings.csv")
hotels

# A tibble: 119,390 × 32
   hotel  is_ca…¹ lead_…² arriv…³ arriv…⁴ arriv…⁵ arriv…⁶ stays…⁷ stays…⁸ adults
   <chr>    <dbl>   <dbl>   <dbl> <chr>     <dbl>   <dbl>   <dbl>   <dbl>  <dbl>
 1 Resor…       0     342    2015 July         27       1       0       0      2
 2 Resor…       0     737    2015 July         27       1       0       0      2
 3 Resor…       0       7    2015 July         27       1       0       1      1
 4 Resor…       0      13    2015 July         27       1       0       1      1
 5 Resor…       0      14    2015 July         27       1       0       2      2
 6 Resor…       0      14    2015 July         27       1       0       2      2
 7 Resor…       0       0    2015 July         27       1       0       2      2
 8 Resor…       0       9    2015 July         27       1       0       2      2
 9 Resor…       1      85    2015 July         27       1       0       3      2
10 Resor…       1      75    2015 July         27       1       0       3      2
# … with 119,380 more rows, 22 more variables: children <dbl>, babies <dbl>,
#   meal <chr>, country <chr>, market_segment <chr>,
#   distribution_channel <chr>, is_repeated_guest <dbl>,
#   previous_cancellations <dbl>, previous_bookings_not_canceled <dbl>,
#   reserved_room_type <chr>, assigned_room_type <chr>, booking_changes <dbl>,
#   deposit_type <chr>, agent <chr>, company <chr>, days_in_waiting_list <dbl>,
#   customer_type <chr>, adr <dbl>, required_car_parking_spaces <dbl>, …
# ℹ Use `print(n = ...)` to see more rows, and `colnames()` to see all variable names

Transformations

After reading in the data, I made a few transformations:

Code

hotels<-hotels %>% 
  rename(is_cancelled = is_canceled) %>%  ## i'm petty
  mutate(booking_dummy = case_when( ## dummy var for whether or not changes were made
    booking_changes == 0 ~ 0,
    T ~ 1
  )) %>% 
  mutate(arrival_date_month_num = case_when( ## numerical variable for months
    arrival_date_month == "January" ~ 1,
    arrival_date_month == "February" ~ 2,
    arrival_date_month == "March" ~ 3,
    arrival_date_month == "April" ~ 4,
    arrival_date_month == "May" ~ 5,
    arrival_date_month == "June" ~ 6,
    arrival_date_month == "July" ~ 7,
    arrival_date_month == "August" ~ 8,
    arrival_date_month == "September" ~ 9,
    arrival_date_month == "October" ~ 10,
    arrival_date_month == "November" ~ 11,
    arrival_date_month == "December" ~ 12
  ))

I found that this data set only required fairly minimal/minor transformations. Firstly, I renamed the “is_canceled” variable to “is_cancelled” primarily because I’m petty, and I knew I would want to spell it with the double “L” the whole time. Next, I created a dummy variable called “booking_dummy” for whether or not any changes were made to a booking, regardless of how many such changes there were. Lastly, I created a variable named “arrival_date_month_num” to assign the corresponding number to each month as named in the “arrival_date_month” column.

Describing the data

Before I started summarising, I wanted to get a sense of what the data “looked like,” so to speak:

Code

names(hotels)

 [1] "hotel"                          "is_cancelled"                  
 [3] "lead_time"                      "arrival_date_year"             
 [5] "arrival_date_month"             "arrival_date_week_number"      
 [7] "arrival_date_day_of_month"      "stays_in_weekend_nights"       
 [9] "stays_in_week_nights"           "adults"                        
[11] "children"                       "babies"                        
[13] "meal"                           "country"                       
[15] "market_segment"                 "distribution_channel"          
[17] "is_repeated_guest"              "previous_cancellations"        
[19] "previous_bookings_not_canceled" "reserved_room_type"            
[21] "assigned_room_type"             "booking_changes"               
[23] "deposit_type"                   "agent"                         
[25] "company"                        "days_in_waiting_list"          
[27] "customer_type"                  "adr"                           
[29] "required_car_parking_spaces"    "total_of_special_requests"     
[31] "reservation_status"             "reservation_status_date"       
[33] "booking_dummy"                  "arrival_date_month_num"        

Code

count(hotels)

# A tibble: 1 × 1
       n
   <int>
1 119390

Code

count(hotels,hotel)

# A tibble: 2 × 2
  hotel            n
  <chr>        <int>
1 City Hotel   79330
2 Resort Hotel 40060

Code

count(hotels,country)

# A tibble: 178 × 2
   country     n
   <chr>   <int>
 1 ABW         2
 2 AGO       362
 3 AIA         1
 4 ALB        12
 5 AND         7
 6 ARE        51
 7 ARG       214
 8 ARM         8
 9 ASM         1
10 ATA         2
# … with 168 more rows
# ℹ Use `print(n = ...)` to see more rows

Code

arrivals<-xtabs(~arrival_date_year+arrival_date_month_num,hotels)
arrivals

                 arrival_date_month_num
arrival_date_year    1    2    3    4    5    6    7    8    9   10   11   12
             2015    0    0    0    0    0    0 2776 3889 5114 4957 2340 2920
             2016 2248 3891 4824 5428 5478 5292 4572 5063 5394 6203 4454 3860
             2017 3681 4177 4970 5661 6313 5647 5313 4925    0    0    0    0

This data set appears to describe hotel bookings from across a number of different countries and kinds of hotels, in the time range between July 2015 and August 2018. Each observation appears to be a single booking, with a range of information about each booking tracked in each column.

Provide Grouped Summary Statistics

As I looked through the data, I found myself gravitating toward the information about booking changes and cancellations.

Code

hotels %>% 
  group_by(is_cancelled) %>% 
  summarise(mean(booking_changes,na.rm=T))

# A tibble: 2 × 2
  is_cancelled `mean(booking_changes, na.rm = T)`
         <dbl>                              <dbl>
1            0                             0.293 
2            1                             0.0983

Code

hotels %>% 
  group_by(is_cancelled) %>%
  summarise(median(booking_changes))

# A tibble: 2 × 2
  is_cancelled `median(booking_changes)`
         <dbl>                     <dbl>
1            0                         0
2            1                         0

Clearly, most bookings did not have any changes to their booking, demonstrated by the fact that the median number of changes for both cancelled and non-cancelled bookings was 0. However, it is interesting to observe that non-cancelled bookings tended to have more booking changes performed, suggesting that making changes to a booking may have increased the likelihood of that booking not having to be cancelled – a boon for hotels and travel agencies, if that extra bit of work is all it takes to retain a customer.

Visualizing and Interpreting

These observations can be seen much more clearly in a graphic visualization:

Code

ggplot(hotels,aes(x=factor(booking_dummy),fill=factor(is_cancelled))) +
  geom_bar() +
  theme_bw() +
  labs(title = "Cancellations by changes in booking", x = "Change in booking?", y= "Number of bookings",
       fill = "Booking cancelled?")

This bar graph utilizes the “booking_dummy” variable, easily dividing all bookings into those that had changes performed and those that didn’t. As can be seen, there were more bookings cancelled without changes being made than there were total bookings with changes! Additionally, a much smaller proportion of bookings with changes ended up being cancelled when compared to bookings without any changes made (though cancellations accounted for less than half of each group in the end). Of course, it is hard to make broad generalizations without knowing more of the story told by this data. However, on the face of things it looks as though a case could be made for flexibility with regards to changing bookings as a strong protection against customers cancelling reservations.