1.

Title: Characteristics of AirBNB Listings and

Hosts in Hong Kong
(code and notes written by Lui Siu Man, MStat 2016-17, please acknowledge if used)

2.Abstract:
In this project, a MS Access database with relational database tables, queries, forms and reports
are created for investigating the characteristics of the listings and hosts of AirBNB in Hong
Kong. In particular, this project would like to identify the AirBNB listing which prices are much
higher than the average price of the listing in each district. Also, this project would like to
identify the hosts who have full marks for all review ratings and what are their listings.

3.Motivation
AirBNB is a company which embrace the share economic idea to allow people to list their
houses, apartments, or rooms on Airbnb.com website, and travelers or anyone need an
accommodation for short stay can search and select the accommodation using the website. Many
think that AirBNB will damage the hotel industry [3] . This project will look at the
characteristics of the AirBNB listings in HK and some summary statistics of the hosts.

4.Model
In order to store the AirBNB listings in a relational database. 6 tables are created in MS Access.
Figure 11 shows the entity-relationship diagram of the database. The 6 tables included are host,
listing, room_type, property_type, bed_type and district.

Figure 1 Booth worksheet output of different staff allocations

5.Data
The dataset is obtained from
http://data.insideairbnb.com/china/hk/hong-kong/2016-08-07/data/listings.csv.gz. This dataset
contains 6424 listings of Airbnb units in Hong Kong. These listings are listed by 3292 different
hosts. The downloaded dataset is a single excel file. The structure of the data has been re-
designed to fit in a relational database schema as shown in Figure 11. That is, normalization has
to been done and primary keys are to setup correctly to link up each database table.

The columns, data types and sizes of each table in the database are listed below:
Table: bed_type
Name Type Size
Bed Type ID Long Integer 4
Bed Type Short Text 25
Table: district
Name Type Size
District ID Long Integer 4
District Short Text 255
Table: host
Name Type Size
Host ID Long Integer 4
Host Name Short Text 255
Host Since Date With Time 8
Response Rate Long Integer 4
Acceptance Rate Long Integer 4
ID Verified Yes/No 1

Table: property_type
Name Type Size
Property Type ID Long Integer 4
Property Type Short Text 255

Table: room_type Page: 8

Name Type Size
Room Type ID Long Integer 4
Room Type Short Text 255
Table: listing
Name Type Size
ListingID Long Integer 4
Title Short Text 255
Description Long Text -
Accommodates Double 8
Bathrooms Double 8
Bedrooms Double 8
Beds Double 8
Price Long Integer 4
Deposit Long Integer 4
Cleaning Fee Long Integer 4
Extra Person Cost Long Integer 4
Minimum Nights Integer 2
Maximum Nights Integer 2
Cleanliness Rating Double 8
Location Rating Double 8
Value For Money Double 8
Host ID Long Integer 4
Property Type ID Long Integer 4
Room Type ID Long Integer 4
Bed Type ID Long Integer 4
District ID Long Integer 4
6.Methodology
To investigate the characteristics of the AirBNB listings, 7 SQL queries are created. Listings in
AirBNB can be classified into different 18 different property types, such as apartment, house,
hut, bed & breakfast, etc. The first query aims to find out the number of listing of different types
of property in each district.
The SQL of the query is :
TRANSFORM Round(Avg(Host_Listing_Details.[Price]),0) AS PriceOfAvg
SELECT Host_Listing_Details.District
FROM Host_Listing_Details
GROUP BY Host_Listing_Details.District
ORDER BY Host_Listing_Details.District
PIVOT Host_Listing_Details.[Property Type];

Figure 12 shows part of the query output. The results show that Yau Tsim Mong district has the
highest number of AirBNB listings, there are 2650 listings in total in Yai Tsim Mong. Most of
the AirBNB listing in Hong Kong are apartments.

Figure 2 Different types of AirBNB property listed in each district

To find out the average price of the listing in each district, the following query is created:
TRANSFORM Round(Avg(Host_Listing_Details.[Price]),0) AS PriceOfAvg
SELECT Host_Listing_Details.District
FROM Host_Listing_Details
GROUP BY Host_Listing_Details.District
ORDER BY Host_Listing_Details.District
PIVOT Host_Listing_Details.[Property Type];

Figure 13shows part of the query output. The results show for apartment, the average prices of
the listings in Tsuen Wan is the highest. While for house, the average prices of the listing in
Central & Western district is the highest.
 Figure 3 Average price of AirBNB property listed in each district

Basic statistics about the prices of listings in each district are shown in Figure 14. The query to
produce this output is:

SELECT DISTINCTROW district.District, Round(Avg(listing.Price),0) AS [Avg Of Price],

Round(Min(listing.Price),0) AS [Min Of Price], Round(Max(listing.Price),0) AS [Max Of Price],
Round(StDev(listing.Price),0) AS [SD Of Price], district.[District ID], Count(listing.ListingID) AS [Count Of
Listing]
FROM district INNER JOIN listing ON district.[District ID] = listing.[District ID]
GROUP BY district.District, district.[District ID]
ORDER BY Round(StDev(listing.Price),0) DESC;

Figure 4 Price statistics of AirBNB property listed in each district

It is also interested to find out the number of listings each host has in each district. The query for
this is:
TRANSFORM Count(Host_Listing_Details.[Response Rate]) AS [Response RateOfCount]
SELECT Host_Listing_Details.[Host ID], Host_Listing_Details.[Host Name], Host_Listing_Details.[Host Since],
Count(Host_Listing_Details.[ListingID]) AS [Count Of Listing]
FROM Host_Listing_Details
GROUP BY Host_Listing_Details.[Host ID], Host_Listing_Details.[Host Name], Host_Listing_Details.[Host Since]
PIVOT Host_Listing_Details.[District];

Figure 15 shows part of the results of the above query. Interestingly, there is a host who have a total of 81
listings in various districts including Wai Chai, Yau Tsim Mong, and Central & Western districts. It is rather
high number for one individual to list so many properties in AirBNB.

Figure 5 Number of listings of each host in each district

Figure 6 Host Form for data input and data viewing

To facilitate data entry and browsing the database, 2 forms are created and linked together. In the
Host from, users can browse the details of each host and the related listings of each host.

7.Analysis
Two reports are created to provide details of the top price listings and top rating host. For top
price listings, they are listings which price of at least 3 standard deviations higher than the
average price of all listings in a district. All together, there are 45 listings are top price listings.
The query used to filter out these top price listings is :
SELECT District_Price_Stats.District, Host_Listing_Details.*
FROM (District_Price_Stats INNER JOIN listing ON District_Price_Stats.[District ID] = listing.[District
ID]) INNER JOIN Host_Listing_Details ON listing.ListingID = Host_Listing_Details.ListingID
WHERE (((Host_Listing_Details.Price)>[Avg Of Price]+3*[SD Of Price]));

Using the above query, a report is created. The report presents the top price listings in each
district and the cleanliness, location, and value for money ratings for each listing. A screenshot
of the first few entries of the report is shown in Figure 17.

Figure 7 High price listings in each district.

Similar, the following query is used to find out the top host whose have more than 3 listings and
their listings have all received full scores for the review ratings from customers. The SQL of the
query is:
SELECT DISTINCTROW host.[Host ID], host.[Host Name], host.[Host Since], host.[Response Rate], host.
[Acceptance Rate], host.[ID Verified], First(listing.ListingID) AS [First Of ListingID], Avg(listing.
[Cleanliness Rating]) AS [Avg Of Cleanliness Rating], Avg(listing.[Location Rating]) AS [Avg Of Location
Rating], Avg(listing.[Value For Money]) AS [Avg Of Value For Money], Count(listing.ListingID) AS
ListingIDOfCount
FROM host INNER JOIN listing ON host.[Host ID] = listing.[Host ID]
GROUP BY host.[Host ID], host.[Host Name], host.[Host Since], host.[Response Rate], host.[Acceptance
Rate], host.[ID Verified]
HAVING (((Avg(listing.[Cleanliness Rating]))=10) AND ((Avg(listing.[Location Rating]))=10) AND
((Avg(listing.[Value For Money]))=10) AND ((Count(listing.ListingID))>3));

Figure 18 shows the results of the above query. There are 4 hosts have more than 3 listings and
have obtained full score (10 out of 10) ratings for all the listings they have.
 Figure 8 Top rated hosts and their listings

8.Discussion
In this project, the characteristics of AirBNB listings and hosts in Hong Kong are examined. Tsim Mong
district are the most popular district of AirBNB listings. Most of the listing are apartments, and it is rather
surprising to see there are some a few house listings. In terms of proportion, the Islands district has the
highest proportion of listings have the house property type. 43 out of 282 listings are of the house
property among the listings in the Islands district. It is also interesting to found that there are 27 hosts
who have at least 20 properties listing in AirBNB.

9.References
[3] Zervas, G., D. Proserpio, and J.W. Byers, The rise of the sharing economy: Estimating the
impact of Airbnb on the hotel industry. Journal of Marketing Research, 2014.

10. Code and file:

Notice:
All the tables, forms, queries, and reports in this project are created and code by myself.
No codes or tables are from others project or other reference sources.

File submitted: airbnb.accdb (MS Access database created by MS office 2016), the database include:
 6 tables
 7 queries
 2 forms
 2 reports