IS428 2016-17 Term1 Assign1 Teo Hui Min

Abstract

The focus of this assignment will be on understanding the private residential property market of Singapore in year 2015 and the purchasing patterns of Singapore residents. I will be identifying possible reasons behind the trend we see from the visualisations, such as the ‘hottest’ regions among Singapore residents.

Problem & Motivation

In the years to come, will people still be able to afford housing? Through this assignment, I would like to find out how the changes in property prices throughout the year has affected the purchasing power of the Singapore residents. Also, finding out some possible factors that will entice people to make a purchase. The main variables that I will be looking at is the average unit price of a property and the number of units sold to understand the purchasers.

Approaches

Data set

Project: The ‘Project’ dataset was used to find out the number of units that were sold in every quarter of the year. It was also used in the assignment to find out the total number of units for a property project, cumulative sold, unsold, unlaunched, launched, completed and uncompleted units. With this data, it will be possible to find out the vacancy and occupancy rate of a project, which will be shown in one of the visualisations.

Transaction: The ‘Transaction’ dataset records property where caveat was lodged after the option-to-purchase was exercised or purchase agreement was signed. The dataset was used to find insights on the property prices, the type of sale, type of property and the planning area and region which the property was built.

Data Exploration

Initially when looking at the datasets, I thought that ‘Transaction’ was solely the number of units sold. However when I compared it to the ‘Project’ datasets, it actually did not tally. An example is the 26 Newton project. <image 9><10> I thought the data was not clean and when online to do a check on the development project. However the information provided online was the same as the downloaded datasets. <image 11><12> Then I realised that the datasets were slightly different.

Transaction dataset: Are transactions with caveats lodged with SLA
Project dataset: Units sold and launched by developers

Data Preparation

As the ‘Project’ datasets which detailed the number of units sold quarterly, the first thing that was done was to combine all the datasets into one file. Based on the file, the data was imported into Tableau for data exploration purposes. Simple drag and drop was performed in Tableau to find out the total number of units that was sold and number of units sold for a particular project. The penetration rate was then calculated and data was exported from Tableau to .csv to be visualised in QGIS. <image 1> In QGIS, a ‘join’ was performed on the csv and planning area SHP file so as to visualise into a choropleth map. <image 2>

To find out the proportion of the units that were sold, unsold and not launched, the data needs to be explored. Through the data exploration, I have mapped out how those measures can be calculated with the columns in the ‘Project’ dataset. <image 3> The measures could be calculated in Tableau, however I have done the calculation (by summing the respective columns) in the dataset first before importing it into Tableau. <image 4>

Other than the above, the rest of the visualisation was performed in Tableau with the datasets.

Tools Utilized

Tableau: Used for data exploration, to understand the data and trends. To visualise the other distribution graphs/charts such as box plot and bar chart.
QGIS: To prepare a choropleth map to visualise the penetration rate in terms of the number of units sold across Singapore.

Results

Penetration Rate

<image 5> Finding 1 (share of private properties supply)
The above choropleth map was to identify the areas with high penetration rate based on the number of units sold across Singapore. North and North-East region experienced the highest penetration rate, and are planning areas such as Yishun (10.46%), Serangoon (7.13%) and Sengkang (20.25).

Finding 2 (share of private properties supply)
The ‘hottest’ selling projects in each of the planning areas are North Park residences, High Park residences and Botanique at Bartley. One similarity between these 3 projects is that all 3 projects are all New Sale projects, which means that new housing may be a more popular choice among the people. By looking at the proportion of units that are sold, all of the projects are at least 65% sold, although it was only launched in the year.

Distribution of property price

By Planning Area Next, let’s focus on the 3 most popular planning areas that we have identified. I would like to find out if the prices of the property there are generally cheaper, which attracted a large number of purchase. Since the property that were developed by the 3 projects were only Apartment, Condominium and Semi-detached house, we will be looking at the average unit price of the property ($psm) across the other planning areas and make a comparison with the median average unit price.

<image 6> Finding 1 (distribution on private properties prices)
Indeed, the average unit price of property in Yishun, Sengkang and Serangoon (highlighted in red) are below or just slightly higher than the median, which means that property in those planning areas are generally cheaper. On the other hand property in the Central region like Orchard and Downtown Core are generally higher in price.

By Type of Sale

Policy Recommendations

Infographics

Improvement