Busy

The first quesiton came into my mind is - are the properties on South Aegean still available? The easiest way to check this is to aggregate calendar dataset by available:

# availability ratio
calendar %>%
  group_by(available) %>%
  summarise(count = n()) %>%
  ggplot(aes(x = available, y = count)) +
  geom_bar(stat = "identity") +
  labs(title="Availability")

Well, many of them are already booked, but more than half is still available.

My next quesiton is - how busy these properties are? This information is useful to avoid the crowds. Here I created a temporary variable busy which represents the number of properties that are already booked:

# current status of busy
calendar %>%
  filter(available == FALSE) %>%
  group_by(date) %>%
  summarise(busy = n()) %>%
  ggplot(aes(x = date, y = busy)) +
  geom_line() +
  scale_x_date(breaks = function(x) seq.Date(from = min(x), to = max(x), by = "1 month")) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  labs(title="Airbnb South Aegean Calendar")

# which day in a week is more busy?
calendar %>%
  mutate(weekday = weekdays(date)) %>%
  filter(available == FALSE) %>%
  group_by(weekday) %>%
  summarise(busy = n()) %>%
  mutate(weekday = fct_relevel(weekday, levels = c("Monday", "Tuesday", "Wednesday", "Thursday", "Friday",  "Saturday", "Sunday"))) %>%
  ggplot(aes(x = weekday, y = busy)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  labs(title="Airbnb South Aegean Calendar") +
  xlab("the day in a week")

The line chart shows that it’s relatively quiet on September and October, which makes sense as it’s typical low travel season at Greece. Starting from November it becomes busier, and there’s a peak at the end of the year - which aligns with the Christmas/New Year holiday season. There is also another peak in March - maybe a festival or national holiday - not so sure about it though :p

The bar chart shows how busy it is in the days of a week. On Friday it’s slightly busier than other days. Avoid Friday, avoid the crowds (a little bit).

Price

Next, the most important factor for any of my travels - price. Let’s see how the price goes with the months in a year:

# which month has highest price?
calendar %>%
  mutate(month = months(date)) %>%
  group_by(month) %>%
  summarise(m_price = mean(price_numeric, na.rm=TRUE)) %>%
  mutate(month = fct_relevel(month, levels = month.name)) %>%
  ggplot(aes(x = month, y = m_price)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  labs(title="Airbnb South Aegean Price") +
  ylab("average price")

Apparently the Airbnb landlord are happier during July and August as they certainly make more money out of this period. The price then drops significantly on September, and again (even more!) on October. My favorite novelist Haruki Murakami wrote in one of his book that “The best period to visit Greece is Spetember and October - if you want to save money.” I’m glad my analysis with Airbnb dataset agrees with him :)

Listing items

Let’s explore more on listing dataset, I mean, there’re 107 attributes for each property after all!

Everyone knows Greece has many islands (1200 - 6000). While this dataset only deal with the islands on South Aegean, still I want to know - which island has more properties on Airbnb:

# which island has more airbnb
listing %>%
  group_by(neighbourhood_cleansed) %>%
  summarise(nr_listing = n()) %>%
  ggplot(aes(x = reorder(neighbourhood_cleansed, -nr_listing), y = nr_listing)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1), axis.title.x = element_blank()) +
  labs(title="Airbnb South Aegean Property") +
  ylab("property count")

Well, I wouldn’t say I’m suprised that Santorini and Mykonos take the first two places on the list. I mean, they are so called ‘hotspot’ in any Greece travel guide. Nevertheless, what surpises me is the third place - Rhodes. While I was there for three days, I was not aware of the fact that it’s also a popular travel destination. I thought the third place might be Kos, but I was very wrong. That’s exactly the reason we should see things in data perspective :)

Then I’m curious about how customers are satisfied with Airbnb properties:

# rating distribution
listing %>%
  ggplot(aes(x = review_scores_rating)) +
  geom_histogram(aes(y = ..density..), binwidth = 1, alpha = 0.9) +
  geom_density(fill = "#FF6666", alpha = 0.2) +
  theme(axis.text.y = element_blank(), axis.title.y = element_blank()) +
  labs(title="Airbnb South Aegean Property") +
  xlab("rating score")

The distribution is very left-skewed - As I expected - most customers are being nice. I think when people travel, they’re more likely to be generous to everything - as long as it does its job.

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##       0      60     100     171     200     999     983

# price distribution for each region
listing %>%
  group_by(neighbourhood_cleansed) %>%
  summarise(nr_listing = n()) %>%
  ungroup() %>%
  right_join(listing, by = c("neighbourhood_cleansed")) %>%
  ggplot(aes(x = reorder(neighbourhood_cleansed, -nr_listing), y = price_numeric)) +
  geom_boxplot() +
  geom_violin(size = 0.2, color = "gray", fill = "gray", alpha = 0.4) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1), axis.title.x = element_blank()) +
  labs(title="Airbnb South Aegean Price") +
  ylab("price")

# stacked histogram for room type
listing %>%
  ggplot(aes(x = price_numeric, fill = room_type)) +
  geom_histogram(position = "stack", binwidth = 10) +
  labs(title="Airbnb South Aegean Property") +
  xlab("price")