"Property Market in Perth, Western Australia"

Reported by Ajeethaa Lüscher
Published on 4 September 2020

Table of Contents

Introduction

This project explores the property market in Perth, Western Australia to help a potential investor in Switzerland who wishes to purchase a property in Perth. Perth is the capital city, also known as the most isolated city in the world according to Bloomberg. It is Australia’s fourth most populous city with an estimated population standing at about 2 085 973 and will continue to grow. The property prices in Perth have been volatile in the past 5 years. However, the demand for units and houses has always stayed steady.

With this idea in mind, an investor approached for a review of property prices and housing developments in Perth. The investor plans to migrate to Perth and is on the lookout for a property in which she could live, although she does not have a fixed envision of a property type or budget. She has never invested in Western Australia and would like a report that would highlight the critical details on housing market there as well on the current and future developments within Perth, before she meets a decision.

To assist the investor, this analysis will examine the historical data on housing prices in Perth city and analyse the surrounding infrastructures which includes restaurants, cafes, and bakeries. The analysis will propose the price range based on property type and location within the city.

Data Procurement

As stated earlier, this is an analysis on the historical housing prices in Perth from 2015 to 2020. Instead of using data from the office of statistics, the data is scraped from the website, reiwa.com.au, a popular real estate portal in Australia, using tools like BeautifulSoup. The prices are then compared based on key variables such as the number of bedrooms, bathrooms, carports, the types of dwelling and the location of each property.

By using Foursquare geographical location data, it will assist to understand the demographics of Perth city and finding out the proximity of the properties to facilities in the city. For the sake of illustration, details such as locations of properties and prices are reflected on the map alongside the facilities such as restaurants, cafes, and bakeries.

Data Extraction and Cleaning

As stated above, the data presented are extracted from the website, reiwa.com.au, using the web scraping tool ‘BeautifulSoup’. The idea is to obtain as much information about a property to conduct exploratory analysis. So the extraction process involves getting information such as number of bedrooms, bathrooms and carport, date of sale, sold price, address and type of dwelling.

Prices are in Australian Dollar and will remain as such. Each page is scrapped and for the sake of simplicity, the data are collated from year 2015 to 2020.

Upon extraction, the data was cleaned for consistencies, tidiness and quality. For example, the data type for date of sale is changed to the date-time-format to get the day of the week. Only the digits from the ‘address’ column are extracted and converted to integer.

In handling missing values, further research was done to see if those values were indeed missing from the website. Some missing values were because of the cleaning process, while the rest was due to incorrect price reporting on the same website. For example, a property was sold for close to a million but the data did not reflect that because the word ‘million’ was not written in digits but rather in words.

Duplicates are first double checked and then removed. There were duplicate addresses and hence are removed. Some rows had the same address but different house numbers. So these data are retained but just edited for consistencies.

This notebook contains only the visuals. For more on data extraction process including code work including web scrapping, please see the Notebook.

Exploratory Data Analysis

As described earlier, for this analysis these key variables are examined:

  • Prices sold between 2015 and 2020 in Perth city;
  • Number of bedrooms, bathrooms, carports for each property sold;
  • The date of sale;
  • Type of dwelling.

The properties are evaluated based on their dimensions and the types of dwelling. As seen below, there are 6 types of properties and they are ’House’, ‘Apartment’, ‘Unit’, ‘Townhouse’, ‘Duplex’, ‘Off The Plan’. There are subtle differences among them.

An apartment is usually part of a high-rise building while a unit is a landed property and is usually part of a strata plan. A townhouse is a multilevel building on a strata title. In other words, the land is shared by other townhouse occupiers. A house is standalone dwelling with the rights to the building and land. Houses tend to be more expensive than townhouses. A duplex is a residential building that holds two homes that share a common central wall on one land. Finally, the off-the-plan which is simply a property sold as a house even before a visible structure has been constructed-Ref.

Some apartments exist as one massive room apartment with no carport. It is like a room on its own and sold as an apartment unit. These apartments are called as zero-bedroom apartments, also known as studio apartments.

Data Visualisation

The dataset has a total of 1021 observations and 7 variables, of which columns ‘address’ and ‘dwelling’ are categorical values and while the rest (except date_sold) are discrete. The highest sale price is about 7.5 million while the lowest recorded is AUD 155,000. The median price is around AUD 445,000 while the average price is above AUD 520,000. The distribution is slight skewed due to the presence of outliers.

Apart from a property (a house) with 8 bedrooms, 4 bathrooms, average property dimensions are 2 x 1 x 1 i.e. 2 bedrooms, 1 bathroom, 1 carport. An interesting development from the sale of the house is now that on that land now sits an apartment building.

A concise summary of the dataframe.

Descriptive statictics of the extracted data.

The property that has the most number of bedrooms and bathrooms which was sold for 1.8 million in 2016. The house was demolished and now stands a high rise apartment building.

In the last 5 years, a great proportion of the properties sold are apartments, which is about 75% as shown in the plot and breakdown below.

Breakdown of types of dwelling and respective count

Houses: The boxplots show the distribution of prices for each dwelling and dimension. Outliers: Only one house* was sold for about 7.5 million. It is a 3 x 1 x 1 house with 278 metre square while a number sold for more than a million but less than 2 million. The average price sold for a house is some where between 1.3 million and 293,000 dollars with some outliers that sold below the minimum price. Overall, most houses were sold for under a million dollar.

Properties that had 4 bedrooms sold better than the ones with fewer. The median price for a 4-bedroom property was 946,000 while for a 3-bedroom the price was 760,000. However, the interesting part is the significant difference between having a 4-bathroom property versus a 3-bathroom. The median price for the former is about 1.83 million while the latter fetched about 1.1 million.

*To see layout of the house, click here.

Apartments: The average price sold is 413,000 dollars and apartments were sold for between 850,000 to 155,000 dollars. There are more outliers in the data for apartments than for the others combined. The highest price sold is about 3 million dollars.

Summary: There is no outlier spotted for townhouses and duplexes. Thesea are the breakdown of median prices sold for each property:

  • a house is 615,000;
  • an apartment is 413,000;
  • a unit is 280,000;
  • a townhouse is 630,000;
  • a duplex is 793,000.

And so: For someone with a budget of 500,000 Australian dollars, one could venture into buying an apartment or a unit in the city. The choice is endless for those with plenty to give.

The dimension of the property that sold the highest in the past 5 years.

What are the dimensions of the properties sold?

Most Units have either one or no carport. But almost all of them have one bathroom and one or two bedrooms. On average, duplexes have 3 bedrooms, 1 bathroom and 1 carport. Some apartments have more than 2 carports. When the data is examined closer, there are only 2 apartments with 4 carports. Both are actually single level penthouses.

Which apartments have 4 carports?

This time the price will be added as a variable to the table to see the price difference versus the dimensions and types of dwelling.

However, when one breaks down the types of dwelling and their associated dimensions, one can see that apartments with 4-bedrooms, 4-bathrooms and 4-carports have the highest selling price. But a closer look reveals the price difference between having more bathrooms and bedrooms. The median price is under 2 million for a 4-bedroom apartment but over 2 million for a 4-bathroom one.

However, when analyses the breakdown even further, a different pattern reveals. Plot B shows a 2-bathroom and 3-bedroom house sold much higher than a 3-bathroom and 3-bedroom house. And further a house with additional bedroom (else equal) sold lower than the former. It appears that properties with more bathrooms sold higher.

----A----
----B----

The next part of the exploratory is to breakdown the date of sale into days. The ideas is to see if there is any pattern to the sale of properties. Most properties were sold on the weekdays, naturally. One can observe that most apartments were sold on a Friday and most houses were sold on a Monday and Thursday. But overall, there is no correlation between prices and the day of week.

There is no correlation between day of the week and price. The two variables that have high correlation to the price are the number of bedrooms and bathrooms, standing at 0.55 and 0.42 respectively.

Results

To visualise where the properties are on a map and their proximity to facilities like restaurants, cafes, and bakeries, geographical data on Perth Central Business District (CBD) is analysed using Foursquare. Once the credentials are established, the GPS coordinates of Perth CBD is acquired. Perth CBD is in the heart of the city, where St. George’s Terrace begins.

So its location is used as a central for this analysis. Data on nearby restaurants, cafes, and bakeries are downloaded programmatically using geocoder Nominatim and data from Foursquare. With this information, a map is then laid out using Folium. One can then see where the facilities are in CBD, along with the locations of properties including prices sold.

After setting up the Foursquare Developer credentials, coordinates of Perth CBD is obtained. Thereafter, information on restaurants, cafes, and bakeries in the CBD are extracted programmatically and transformed into three separate dataframes.

To search for restaurants within Perth CBD

The radius is set to 1000m within the CBD. The focus is only on restaurants and not the categories. The investor does not have a particular favourite for the choice of cuisine type. The dataframe below shows all the restaurants captured by Foursquare within the CBD.

The dataframe is then filtered to extract the name and categories for mapping later.

To search for cafes within CBD

The dataframe below shows all the cafes captured by Foursquare within the CBD.

The dataframe is then filtered to extract the name and categories for mapping later.

To search for bakeries within CBD

The dataframe below shows all the bakeries captured by Foursquare within the CBD.

The dataframe is then filtered to extract the name and categories for mapping later.

Next is to get the GPS coordinates for all the addresses on the dataframe. The extraction is programmatically and its process can be viewed in the Notebook. The extracted data is then saved as a csv file named coordinates. Some addresses have missing GPD coordinates and so the mapping of the properties is then limited to those addresses with coordinates.

The map of restaurants, cafes, and bakeries in the vicinity of the properties

The map below is prepared using Folium which is interactive. It shows the location of restaurants in blue, cafes in yellow and bakeries in green. The red circles are location of the properties and will display prices when clicked.

Discussion / Recommendations

As the investor has no fixed budget or property type in mind, through this analysis a few recommendations are put forward to her. Since she wishes to buy near business facilities, the map reveals a number of options to filter through the sea of information. There are properties for sale in various locations near the CBD. To assist the investor, the search of properties will be narrowed to these streets; Murray Street, St. Georges Terrace, Wellington Street, Hay Street, and William Street. The dataframe below shows all the properties on those streets. There were about 300 properties that were sold in the last 5 years.

This is the breakdown of properties sold in the CBD area in the past 5 years. The highest price is 1.2 million and the lowest is 184,000.

There were more apartments sold in the CBD than the others.

Using the plot below, one can rule out the option of buying a studio apartment. So the choice is between getting a 1-bathroom and a 2-bathroom apartments as opposed to the others. An average price of a 2-bedroom and 2-bathroom apartment is about 500,000, while for a 3-bedroom and 2-bathroom apartment fetched about 560,000.

However, most apartments sold for under a million dollars except for several apartments as they turned out to be penthouses, which cost more.

Conclusion

Based on the analysis, I would suggest to the investor to have a budget of a million dollars and to narrow her search to purchasing a 3-bedroom and 2-bathroom apartment in the CBD.

Based on the recommendation, I went on reiwa.com.au and filtered the search based on this findings. These are the results:

References