AirBnB User Pathways - Data Description

User pathways are the routes by which people navigate a website. The AirBnB data set contains data on user pathways for user sessions in the past year in a US city. We will be using R to perform data analysis and visualization to explore and identify trends in user pathways, and uncover insights to understand how people are using the AirBnB site.

# Find total number of observations  
nrow(AirB_tib)

# Get variable names
names(AirB_tib)

# Get summary information on the AirBnB data set variables
summary(AirB_tib)

library(tibble) # used to create tibbles
library(tidyr) # used to tidy up data
library(prettydoc) # document themes for R Markdown
library(DT) # used for displaying R data objects (matrices or data frames) as tables on HTML pages
library(lubridate) # date/time functions

Import Data

# Read data via the source data url provided above
AirBnB <- read.delim(url
  ("http://databits.io/static_content/challenges/airbnb-user-pathways-challenge/airbnb_session_data.txt"), 
  sep = "|", na.strings = 'NULL')

# Store data as tibble
AirB_tib <- as_tibble(AirBnB)

# Preview data - show first 8 rows
datatable(head(AirB_tib, 8),options = list(scrollX=TRUE, pageLength=4))

Data Cleaning

First, we would like to separate the column “dim_device_app_combo” into two columns - one for the device name and one for the application name. I want to do the same to the column “next_dim_device_app_combo” as well.

# Separate Device and Application columns

AirB_tib2 <- AirB_tib %>% 
  separate(dim_device_app_combo, 
           into = c("Device", "Application"), sep = " - ") %>%
  separate(next_dim_device_app_combo,
           into = c("Next_Device", "Next_Application"), 
           sep = " - ") 

datatable(head(AirB_tib2, 8),options = list(scrollX=TRUE, pageLength=4))

Next we want to calculate the duration of each session, by creating a new calculated field, which is the difference between the session start time and the session end time. In order to do so, we would first convert the start time and end time strings into date-time (hour/minute/second) format, using the hms() function from the lubridate package.

Then, we would calculate the difference of the start and end times to calculate the duration of the session in a calculated field called “Duration”. We would also calculate the duration of the next session in a calculated field called “Next_Duration”.

# Convert the start and end times from string to date/time format

AirB_tib2$Start_Time <- ymd_hms(AirB_tib2$ts_min)
AirB_tib2$End_Time <- ymd_hms(AirB_tib2$ts_max)

AirB_tib2$Next_Start_Time <- ymd_hms(AirB_tib2$next_ts_min)
AirB_tib2$Next_End_Time <- ymd_hms(AirB_tib2$next_ts_max)

# When we substract the difference of the times, the result is in seconds,
# so we divide by 60 to get the duration in minutes

AirB_tib2$Duration <- 
  (AirB_tib2$End_Time - AirB_tib2$Start_Time) / 60

AirB_tib2$Next_Duration <- 
  (AirB_tib2$Next_End_Time - AirB_tib2$Next_Start_Time) / 60

datatable(head(AirB_tib2, 8),options = list(scrollX=TRUE, pageLength=4))

Planned Analysis

We will explore various ways of analyzing the AirBnB data set - potential ideas are listed below:

1. Investigate whether there is a correlation between the amount of time spent on the site, and whether the user takes action (as in, performs a search, sends a message, or sends a booking request).

2. Explore whether users are more likely to place a booking request based on the device they are using. Are people more likely to request booking when they are browsing on computers vs. when they are browsing on mobile devices?

3. Explore whether users are more likely to place a booking request based on the application/browser they are using. For instance, are people more likely to place a booking on Chrome vs. on Firefox? This could potentially suggest that the site is easier to navigate on one application vs. another?

4. Examine the actions of the users in each session, and determine whether users tend to visit the site multiple times before requesting a booking. If so, determine approximately how many times users typically visit the site before requesting a booking.

5. Examine the session information by unique visitors and note the behavior of frequent visitors (say 10 sessions or more). How often do frequent visitors spend on the site? Is there a short duration in the number of days between sessions right before placing a booking request?