When thinking about factors that may influence the volume of hotel booking and arriving, time is one of the things that should not be ignored. Because seasonality contributes to a variation in guest flows, and the timing of hotel-booking and cancellation could imply patterns of the consumer's decision making behavior. Some other indexes, such as the length of the lead time, will also impact a consumer's final decision.
Is there a seasonality in the number of hotel bookings?
A trend of seasonality in the number of hotel bookings can be found from the graph. In both City and Resort Hotel, bookings made for April to June and September to October were more than that of the other months. It seems that those hotels were popular during spring and autumn. Whereas bookings between June and August and between November and February decreased, probably because of weather constraints in summer and winter. We can infer that the two hotels are located somewhere with pleasant autumns and springs, when the weather becomes nice and travelers are being attracted. In summer and winter, the climate may become not captivating thus there were sudden decreases in the booking numbers. From the dataset, we can see the City Hotel actually locates in Lisbon, and the resort hotel locates in Algarve, Portugal. Those two cities both featuring Mediterranean climate, which means mild spring and autumn; hot, dry summers with average highs of more than 80 °F, and that may be a main reason of decline in the number of tourists during summers.
When seeing the bookings by arrival week numbers, it can be seen that both hotels were busy from week number 13 to 25 and week number 35 to 42, when spring and autumn took place. This trend corresponds to the tendency in the number of bookings by arrival month chart. The reason for this high number of bookings could be nice weather during April to June and September to October. Many major public holidays also fall during week number 13 to 25, for example Good Friday, Easter, Freedom Day (Apr 25th), Labor Day (May 1st), Corpus Christi, and Portugal Day (Jun 10th), thus the demand of the hotels will increase during those weeks. The number of guests dropped for both hotels in week 51. It was the week of Christmas, and people may want to spend time with their families at home rather than stay in a hotel.
Comparison of length of stay between adult with and without children or babies.
We would like to check if the length of the stay is related to the type of the guests. It can be noticed that more adult guests without kids booked for the two hotels than the number of adults brought children or babies. We can then see that most adults with children or babies tended to stay for 3 days, and most adults alone stayed for 2 days. It can be understood since many people are in the hotel for short-term stays, such as a trip or holiday or business meeting. Fewer people tended to stay for a longer period, since there might be no long conference held nearby, and both of the hotels may not be hotel apartments that suits for a long time living.
From the graph by month, we could see that the distribution of adult guest per month basically suits the overall seasonality: more adults without kids ordered the hotels in spring and autumn. However, the case is different for adults with children or babies: the number of adults brought their kids increased during the summertime and to a peak in August, when summer break took place. When school started, the number of guests with children decreased significantly in September, showing that children were at school. It shows that since most kids have to enter the education system, the school calendar determines their availability rather than other factors. Children and babies are more likely to stay in a hotel during their school breaks no matter what the climate is. Hotels could design certain marketing strategies targeting different groups of guests, for example summer family room discounts for adults with their kids.
What is the reservation status by month?
The dataset we used contains both bookings that actually arrived and booking that were cancelled or not show for the arrival date. When analyzing the composition of the guest status by month, we could see that most guests arrived as planned, and only a small part was absent without cancellation. The check-in rate was the highest while the cancellation rate was the lowest in January. Months with the top3 cancellation rate were June, April, May and September (tie). Which makes sense since spring and autumn are the peak seasons of the hotel, the large guest flow drove the increased number of guests as well as the number of people who changed their bookings. The high not-show rate in February may be caused by the absence of certain travel groups. All the guests booked with three certain travel agents did not show up in Feb 2017, which adds up to 150 guests, probably because of the change of the travel plan or some kind of emergency.
It should be noticed that the month when the action of cancellation took place is not the same as the arrival month being cancelled for (since many guests cancel ahead of their time, for example cancel a room booked for March in January). We could see that January was the month when people made most future cancellations, probably because of a feeling of entering a new year and the need to make plans about time and money again. Each of the Months being highlighted has more cancellations from other months than the number of cancellations made in that month itself. We can see that five out of those six months are with most bookings in consequence, which means more orders during the high seasons were cancelled in advance months, probably because people changed their travel plans. The only exception fell in August, many guests cancelled their trips a head of schedule probably due to the severe weather and would like to go to somewhere else.
Is there a behavioral pattern behind guests' cancellation?
After exploring the rate of cancellation for each month, what came into our mind is if the behavior of cancellation has some kind of pattern. For the guests who cancelled their orders, we find that there is a negative correlation between the number of days before the expected arrival date and the number of guests who cancelled their bookings. We can see that there were more than 880 guests cancelled their bookings on the day they should check-in, and the number of cancellations decreased gradually as the number of days before arriving went up. It can be seen as people's tendency of making decisions about changing or considering about the upcoming event at the last minute. For guests who cancelled their bookings long before their arrival, it can be interpreted as some part of their plans had changed during their long-term of reservation. In a word, plans always fall behind changes, hotel owners should set policy of cancellation accordingly, to best fill in the gap between people who cancel and people who booking at the last minute.
How does the lead time relate to the number of changes made for booking?
Lead time is defined as the number of time between the booking is made and the arrival date. We are curious if a longer lead time could lead to a increased number of booking changes made since people may change their ideas while waiting. For the guests who made changes to their bookings, we can see that most of them had a lead time less than 30 days, whereas the second largest proportion belonged to guests with a lead time greater than 180 days. No matter how long the lead time was, we find that guests in each subgroup mostly made 1 change, and the second most frequent number of changes was 2. It can be concluded that people do not make changes unlimitedly just because they are capable of doing that. Guests may seek chances to adjust to their needs in certain extent, so hotel managers should try to utilize those limited chances to keep those consumers.
Conclusion
1.There is a seasonality in the number of bookings for those two hotels. Both City and Resort hotel had more orders in spring and autumn and less orders in summer and winter during the observed period. The main reason may be holidays and weather.
2.There were more guests stay without children or babies, and most guests booked for short-term stays. The number of guests with kids increased significantly during the months of school break. So it is the school calendar that determines the seasonality of the bookings with children or babies.
3.Those two hotels had a relatively high rate of cancellation during their peak seasons, but more orders were being cancelled in previous months rather than same-month cancellation.
4. Most guests cancelled their bookings on their expected arrival date, and the hotels should set policy of cancellation accordingly.
5.Most guests just made 1 or 2 changes on their bookings regardless of how long the lead time between their order and their arrival was.