In hnguyen1174/online_visits_assessment: To assess the effectiveness of online visit on booking metrics

1. Load Libraries

library(tidyverse)
library(here)
library(lubridate)
library(janitor)
library(zoo)
library(ggthemes)

library(forecast)
library(stargazer)
library(data.table)
library(tableone)
library(lattice)
library(pwr)
library(rcompanion)
library(scales)
library(plm)
library(readxl)
library(MatchIt)
library(lfe)
library(Synth)
library(gsynth)
library(panelView)
library(CausalImpact)

devtools::load_all()

2. Load and Process Data

metrics <- readr::read_csv(file.path(here::here(), 'data/data.csv'))
property <- readr::read_csv(file.path(here::here(), 'data/property_info.csv'))

c(metrics, property) %<-% get_property_data()

property_FL <- property %>% 
  filter(location == 'Destin, Florida')

property_TX <- property %>% 
  filter(location == 'Austin, Texas')

property_NV <- property %>% 
  filter(location == 'Vegas, Nevada')

3. Exploratory Data Analysis - Properties

3.1. Number of Properties by Location and Online Visit Status

p <- property %>% 
  group_by(location, online_visits) %>% 
  tally() %>% 
  ungroup() %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = location, y = n, fill = `Online Visits`, label = n)) +
  geom_col(position = 'dodge') +
  geom_text(position = position_dodge(width = 0.9), vjust = -1) +
  ylim(0, 650) +
  ylab('Location') +
  xlab('Number of Property') +
  theme_hc() +
  ggtitle('Number of Properties by Location')

print(p)
ggsave(file.path(here::here(), 'fig/01_num_prop_location.png'), p)

3.2. Mean Ratings and Mean Number of Reviews by Online Visit Status

p <- property %>% 
  group_by(online_visits) %>% 
  summarize(mean_rating = mean(avg_review_rating_num),
            mean_num_reviews = mean(number_of_reviews_on_the_property_num)) %>% 
  ungroup() %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  gather(key = 'metrics', value = 'value', -`Online Visits`) %>% 
  mutate(metrics = if_else(metrics == 'mean_rating', "Average Rating", "Average Number of Rating")) %>% 
  ggplot(aes(x = metrics, y = value, fill = `Online Visits`, label = round(value, 2))) +
  geom_col(position = 'dodge') +
  geom_text(position = position_dodge(width = 0.9), vjust = -1) +
  xlab('Metrics') +
  ylab('Value') +
  theme_hc() +
  ggtitle('Mean Ratings and Mean Number of Reviews by Online Visit Status') +
  ylim(0, 15)

print(p)
ggsave(file.path(here::here(), 'fig/02_avg_rating_review_by_tour.png'), p)

p <- property %>% 
  group_by(location, online_visits) %>% 
  summarize(mean_rating = mean(avg_review_rating_num),
            mean_num_reviews = mean(number_of_reviews_on_the_property_num)) %>% 
  ungroup() %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  gather(key = 'metrics', value = 'value', -`Online Visits`, -location) %>% 
  mutate(metrics = if_else(metrics == 'mean_rating', "Average Rating", "Average Number of Rating")) %>% 
  ggplot(aes(x = metrics, y = value, fill = `Online Visits`, label = round(value, 2))) +
  geom_col(position = 'dodge') +
  geom_text(position = position_dodge(width = 0.9), hjust = -0.15) +
  xlab('Metrics') +
  ylab('Value') +
  theme_hc() +
  ggtitle('Mean Ratings and Mean Number of Reviews by Online Visit Status') +
  ylim(0, 20) +
  coord_flip() +
  facet_wrap(~location)

print(p)
ggsave(file.path(here::here(), 'fig/02B_avg_rating_review_by_tour_by_loc.png'), p, width = 8, height = 4, dpi = 300)

There is weak evidence that the properties in the online visit group has higher average ratings than the non-online visit group.

continuous_vars <- c('avg_review_rating_num', 'number_of_reviews_on_the_property_num')
map(continuous_vars, vectorize_ttest_for_property, property)

When breaking into specific locations: There is weak evidence that the properties in the online visit group has higher average ratings than the non-online visit group in Texas. There is weak evidence that the properties in the online visit group has lower average number of reviews than the non-online visit group in Nevada

continuous_vars <- c('avg_review_rating_num', 'number_of_reviews_on_the_property_num')
cat('================= FLORIDA ================= \n\n')
map(continuous_vars, vectorize_ttest_for_property, property_FL)

cat('================= TEXAS ================= \n\n')
map(continuous_vars, vectorize_ttest_for_property, property_TX)

cat('================= NEVADA ================= \n\n')
map(continuous_vars, vectorize_ttest_for_property, property_NV)

3.3. Property Size by Online Visit Status

p <- property %>% 
  group_by(online_visits) %>% 
  summarize(`Avg # of Bedrooms` = mean(bedrooms),
            `Avg # of Bathrooms` = mean(bathrooms),
            `Avg Maximum Occupancy` = mean(maximum_occupancy)) %>% 
  ungroup() %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  gather(key = 'metrics', value = 'value', -`Online Visits`) %>% 
  ggplot(aes(x = metrics, y = value, fill = `Online Visits`, label = round(value, 2))) +
  geom_col(position = 'dodge') +
  geom_text(position = position_dodge(width = 0.9), vjust = -1) +
  xlab('Metrics') +
  ylab('Value') +
  theme_hc() +
  ggtitle('Average Property Size by Online Visit Option') +
  ylim(0, 10)

print(p)
ggsave(file.path(here::here(), 'fig/03_prop_size_by_online_visit.png'), p)

p <- property %>% 
  group_by(location, online_visits) %>% 
  summarize(`Avg # of Bedrooms` = mean(bedrooms),
            `Avg # of Bathrooms` = mean(bathrooms),
            `Avg Maximum Occupancy` = mean(maximum_occupancy)) %>% 
  ungroup() %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  gather(key = 'metrics', value = 'value', -`Online Visits`, -location) %>% 
  ggplot(aes(x = metrics, y = value, fill = `Online Visits`, label = round(value, 2))) +
  geom_col(position = 'dodge') +
  geom_text(position = position_dodge(width = 0.9), hjust = -0.15) +
  xlab('Metrics') +
  ylab('Value') +
  theme_hc() +
  ggtitle('Average Property Size by Online Visit Option') +
  ylim(0, 15) +
  coord_flip() +
  facet_wrap(~location)

print(p)
ggsave(file.path(here::here(), 'fig/03B_prop_size_by_online_visit_by_loc.png'), p, width = 8, height = 4, dpi = 300)

With p-values > 0.05, none of these variables are that different between the online visit group and non-online visit group.

continuous_vars <- c('bedrooms', 'bathrooms', 'maximum_occupancy')
map(continuous_vars, vectorize_ttest_for_property, property)

continuous_vars <- c('bedrooms', 'bathrooms', 'maximum_occupancy')

cat('================= FLORIDA ================= \n\n')
map(continuous_vars, vectorize_ttest_for_property, property_FL)

cat('================= TEXAS ================= \n\n')
map(continuous_vars, vectorize_ttest_for_property, property_TX)

cat('================= NEVADA ================= \n\n')
map(continuous_vars, vectorize_ttest_for_property, property_NV)

3.4. Comparisons: Instant bookings and premier partner badge

p <- property %>% 
  group_by(online_visits) %>% 
  summarize(instant_booking_enabled = mean(instant_booking_enabled),
            premier_partner_badge = mean(premier_partner_badge)) %>% 
  ungroup() %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  gather(key = 'metrics', value = 'value', -`Online Visits`) %>% 
  ggplot(aes(x = metrics, y = value, fill = `Online Visits`, label = scales::percent(value, 0.01))) +
  geom_col(position = 'dodge') +
  geom_text(position = position_dodge(width = 0.9), vjust = -1) +
  xlab('Metrics') +
  ylab('Value') +
  theme_hc() +
  ggtitle('Percentage of Properties with Instant Booking and Premier Partner Badge') +
  ylim(0, 0.4)

print(p)
ggsave(file.path(here::here(), 'fig/04_IB_and PP_type_by_tour.png'), p)

p <- property %>% 
  group_by(location, online_visits) %>% 
  summarize(instant_booking_enabled = mean(instant_booking_enabled),
            premier_partner_badge = mean(premier_partner_badge)) %>% 
  ungroup() %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  gather(key = 'metrics', value = 'value', -`Online Visits`, -location) %>% 
  ggplot(aes(x = metrics, y = value, fill = `Online Visits`, label = scales::percent(value, 0.01))) +
  geom_col(position = 'dodge') +
  geom_text(position = position_dodge(width = 0.9), hjust = -0.15) +
  xlab('Metrics') +
  ylab('Value') +
  theme_hc() +
  ggtitle('Percentage of Properties with Instant Booking and Premier Partner Badge') +
  ylim(0, 0.6) +
  coord_flip() +
  facet_wrap(~location)

print(p)
ggsave(file.path(here::here(), 'fig/04B_IB_and PP_type_by_tour_by_loc.png'), p, width = 10, height = 4, dpi = 300)

With p-values > 0.05, instant_booking_enabled and premier_partner_badge are that different between the online visit group and non-online visit group.

indicator_vars <- c('instant_booking_enabled', 'premier_partner_badge')
map(indicator_vars, vectorize_propztest_for_property, property)

There is weak evidence (p-value < 0.1) that there are more more properties with instant booking enabled with the online visit option than without the non-online visit option in Texas.

indicator_vars <- c('instant_booking_enabled', 'premier_partner_badge')

cat('================= FLORIDA ================= \n\n')
map(indicator_vars, vectorize_propztest_for_property, property_FL)

cat('================= TEXAS ================= \n\n')
map(indicator_vars, vectorize_propztest_for_property, property_TX)

cat('================= NEVADA ================= \n\n')
map(indicator_vars, vectorize_propztest_for_property, property_NV)

3.5. Number of Property Types by Online Visit Option

p <- property %>% 
  group_by(property_type, online_visits) %>% 
  tally() %>% 
  ungroup() %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = reorder(property_type, n), y = n, fill = `Online Visits`)) +
  geom_col(position = 'dodge') +
  coord_flip() +
  theme_hc() +
  ylab('Number of Property') +
  xlab('Type') +
  ggtitle('Number of Property by Type by Online Visit Option')

print(p)
ggsave(file.path(here::here(), 'fig/05_number_prop_by_type.png'), p)

p <- property %>% 
  group_by(location, property_type, online_visits) %>% 
  tally() %>% 
  ungroup() %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = reorder(property_type, n), y = n, fill = `Online Visits`)) +
  geom_col(position = 'dodge') +
  coord_flip() +
  theme_hc() +
  ylab('Number of Property') +
  xlab('Type') +
  ggtitle('Number of Property by Type by Online Visit Option') +
  facet_wrap(~location)

print(p)
ggsave(file.path(here::here(), 'fig/05B_number_prop_by_type_by_loc.png'), p, width = 8, height = 4, dpi = 300)

4. Exploratory Data Analysis - Metrics

4.1. No Location (Aggregated)

metrics_by_months <- metrics %>% 
  left_join(property, by = 'property_id') %>% 
  mutate(year_month = as.yearmon(date)) %>% 
  group_by(date, online_visits) %>% 
  summarize(bookings_per_prop = mean(bookings, na.rm = TRUE),
            booking_amount_per_prop = mean(booking_amount, na.rm = TRUE),
            booking_requests_per_prop = mean(booking_requests, na.rm = TRUE),
            property_page_views_per_visit_per_prop = mean(property_page_views_visits, na.rm = TRUE),
            displays_in_searchpage_per_prop = mean(number_of_times_the_property_got_displayed_in_search_page, na.rm = TRUE)) %>% 
  ungroup()

p <- metrics_by_months %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = booking_amount_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Booking Amount ($) Per Property') +
  ggtitle('Average Booking Amount Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc()

print(p)
ggsave(file.path(here::here(), 'fig/06A_booking_amount_per_prop_nolo.png'), p)

p <- metrics_by_months %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = bookings_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Booking Per Property') +
  ggtitle('Average Booking Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc()

print(p)
ggsave(file.path(here::here(), 'fig/06B_bookings_per_prop_nolo.png'), p)

p <- metrics_by_months %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = booking_requests_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Booking Requests Per Property') +
  ggtitle('Average Booking Requests Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc()

print(p)
ggsave(file.path(here::here(), 'fig/06C_booking_requests_per_prop_nolo.png'), p)

p <- metrics_by_months %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = property_page_views_per_visit_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Page Views Per Visit Per Property') +
  ggtitle('Average Page Views Per Visit Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc()

print(p)
ggsave(file.path(here::here(), 'fig/06D_page_views_per_visit_per_prop_nolo.png'), p)

p <- metrics_by_months %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = displays_in_searchpage_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Displays in Search Page Per Property') +
  ggtitle('Average Displays in Search Page Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc()

print(p)
ggsave(file.path(here::here(), 'fig/06E_displays_in_searchpage_per_prop_nolo.png'), p)

4.2. By Location

metrics_by_months_by_loc <- metrics %>% 
  left_join(property, by = 'property_id') %>% 
  mutate(year_month = as.yearmon(date)) %>% 
  group_by(date, online_visits, location) %>% 
  summarize(bookings_per_prop = mean(bookings, na.rm = TRUE),
            booking_amount_per_prop = mean(booking_amount, na.rm = TRUE),
            booking_requests_per_prop = mean(booking_requests, na.rm = TRUE),
            property_page_views_per_visit_per_prop = mean(property_page_views_visits, na.rm = TRUE),
            displays_in_searchpage_per_prop = mean(number_of_times_the_property_got_displayed_in_search_page, na.rm = TRUE)) %>% 
  ungroup()

p <- metrics_by_months_by_loc %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = booking_amount_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Booking Amount ($) Per Property') +
  ggtitle('Average Booking Amount Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc() +
  facet_wrap(~location, nrow = 3) 

print(p)
ggsave(file.path(here::here(), 'fig/06A_booking_amount_per_prop_byloc.png'), p, height = 7, width = 5, dpi = 300)

p <- metrics_by_months_by_loc %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = bookings_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Booking Per Property') +
  ggtitle('Average Booking Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc() +
  facet_wrap(~location, nrow = 3) 

print(p)
ggsave(file.path(here::here(), 'fig/06B_bookings_per_prop_byloc.png'), p, height = 7, width = 5, dpi = 300)

p <- metrics_by_months_by_loc %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = booking_requests_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Booking Requests Per Property') +
  ggtitle('Average Booking Requests Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc() +
  facet_wrap(~location, nrow = 3) 

print(p)
ggsave(file.path(here::here(), 'fig/06C_booking_requests_per_prop_byloc.png'), p, height = 7, width = 5, dpi = 300)

p <- metrics_by_months_by_loc %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = property_page_views_per_visit_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Page Views Per Visit Per Property') +
  ggtitle('Average Page Views Per Visit Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc() +
  facet_wrap(~location, nrow = 3) 

print(p)
ggsave(file.path(here::here(), 'fig/06D_page_views_per_visit_per_prop_byloc.png'), p, height = 7, width = 5, dpi = 300)

p <- metrics_by_months_by_loc %>% 
  dplyr::rename(`Online Visits` = online_visits) %>% 
  ggplot(aes(x = date, y = displays_in_searchpage_per_prop, color = `Online Visits`)) +
  geom_line() +
  geom_smooth(method = 'loess', se = FALSE) +
  xlab('Month') +
  ylab('Average Displays in Search Page Per Property') +
  ggtitle('Average Displays in Search Page Per Property by Month') +
  geom_vline(xintercept = as.Date('2019-01-01'), 
             linetype = 'longdash', 
             color = 'black', 
             size = 0.3) +
  theme_hc() +
  facet_wrap(~location, nrow = 3) 

print(p)
ggsave(file.path(here::here(), 'fig/06E_displays_in_searchpage_per_prop_byloc.png'), p, height = 7, width = 5, dpi = 300)

hnguyen1174/online_visits_assessment documentation built on Dec. 20, 2021, 4:46 p.m.