TROUT1 (Zhixin & Sam & Josh) 2 Dec. 2021
In this project we will explore how factors such as: Location, House type, Area and Number of Bedrooms influence the price of property in London. The goal of our project is to identify which factors have the greatest influence on the price of housing in London.
The dataset we used was obtained from Kaggle. According to the uploader the information was obtained via web scraping. The variables we are going to be focusing on in this project are, Region, Area and House type.
X- ID of the property.
Property.Name- The name of the property.
Price- The price of the property.
House Type- The type of the house.
Area in sq ft- Aera in square feet of the property.
No. of Bedrooms- Number of bedrooms.
No. of Bathdrooms- Failure data column, please ignore.
No. of Receptions- Failure data column, please ignore.
Location- The name of the neighbourhood of the house.
City/County- The city or county the property located.
Postal Code- The postal code of the property
As you can see, a major issue with our dataset was that the values of the variables ‘No. of Bedrooms’, ‘No. of Bathrooms’ and ‘No. of Receptions’ were the same and so could not be used. We found that this data corresponds to the ‘No. of Bedrooms’. As well as this we found that the location variable does not have entries for each listing and that the dataset contains houses from outside of London.
We cleaned our dataset in 5 steps.
-
Filtering out all houses that are not in London.
-
Removing the variables “No. of bathrooms” and “No. of receptions.”
-
Creating variable: “price_per_sq_ft” = “Price”/“Area in sq ft”
-
Extracting the District Code from the “Postal Code” variable.
-
Assigning each District code to one of the 4 large regions.
Dividing Rule:
Before we start our analysis we will expand on how we assigned each property to one of 4 regions. We have chosen to divide the District codes into 4 regions shown on the map below
Different marked colour represents different regions:
-
Red represents group “A-Central”
-
Blue represents group “B-Middle”
-
Red represents group “C-Outer”
-
Unmarked region represents group “D-Further outer”
What we expected to see is that: The closer to the city center, the more expensive the land is. That is, we should observe a higher price per sq ft in areas closer to the center.
We are curious if there is an outlier region in “Further Outer” that has a higher housing price per sq ft than the “Outer” region which is skewing our result. So we made a map to see the distribution of housing prices. It seems that no such region exists. The housing price per sq ft in “Further outer” is higher than “Outer” region in general.
We are curious if there is an outlier region in “Further Outer” that has a very high housing price per sq ft which is skewing our result. So we made a map to see the distribution of housing prices. It seems that no such region exists. The housing price per sq ft in “Further outer” is higher than “Outer” region in general.
Prices grouped by region and type of housing
We thought that maybe the difference in mean price between the outer and further outer regions was concentrated into one housing type. For example, maybe the outer region has a large number of cheaper apartment complexes, pulling its mean housing prices down . However it seems that all popular housing types are on average more expensive in the further outer region.
Table showing correlation between Area and Price over all large regions:
Firstly we did a simple area versus price graph over all regions unfortunately the graph came out feeling cluttered due to severe overplotting. So I instead used glance to create a table to show there is a strong correlation between area and price of housing. We can tell this as the graph has a relatively high adjusted r squared value of 0.575.
Table showing correlation between Area and Price over the Middle region:
Then we filtered the data to only 1 large region. The points seem to follow a much clear correlation; this is proven from an increase in adjusted r squared from 0.575 to 0.730 which shows that there is a much stronger correlation between price and area when there is less regional variance.
Generally as you move further from the center of London house prices and price per square foot decreases. However our data tells us that on average homes in the further outer sector are more expensive than houses in the outer sector this did not fit with the trend and is likely due to the prevalence of cheaper terrace housing in the outer region when compared to the Further Outer region where you may see more suburban detached family homes. There is a simplistic linear positive correlation between the area and price of a home; this can be seen as the linear regression model between the two has a relatively high adjusted r squared value of 0.575 . When the points are plotted on a scatter plot you can also see a clear positive correlation particularly when filtering by region.
The data set was scraped poorly. It had repeated values under the number of bathrooms,bedrooms and reception variables. After removing the number of bathrooms and lobby variables our data set contained a limited amount of data to manipulate. The data set only contained observations from certain regions of London.Of the 82 regions with data points almost half had under 20 data points. This makes our conclusions drawn about how location affects house price less reliable. The data set gave limited information on specifics of the properties for example if the houses were terrace, detached or semi detached. As well as if the homes came with any land or any amenities.
Our presentation can be found in “presentation.html”
Kulkarni, A., 2021. Housing Prices in London. [online] Kaggle.com. https://www.kaggle.com/arnavkulkarni/housing-prices-in-london [Accessed 19 November 2021].
- Data.london.gov.uk. 2021. Ordnance Survey Code-Point Open – London Datastore. [online] Available at: https://data.london.gov.uk/dataset/ordnance-survey-code-point [Accessed 29 November 2021].