Group10 research paper

Proposal

Poster

Application

Research Paper

SPIVA

A Singapore Property Interactive Visual Analytics Dashboard

Abstract – The [something about data source or collection of data] provides us with huge amount of data about prices of private properties in different parts of Singapore. However, these data do not tell us much at a transactional level. [Add on something about what is lacking to the users now]. To overcome this issue, we designed and developed SPIVA, Singapore Property Interactive Visual Analytics Dashboard, to democratise data and analytics by allowing property marketplace stakeholders to explore and analyse Singapore private property data to gain valuable insights about Singapore property market price trends and expectations. We showcase the potential of SPIVA through the use of various visually captivating plots like choropleth, ridge plots, box plots and geo-facets line graph.

Keywords – Singapore Private Property Price Trends, Geo-facet Line Graph, Ridges Plot, Box Plots

I. INTRODUCTION

II. MOTIVATION AND OBJECTIVES

Our research and development efforts were motivated by the general lack of effective and easy to use web-enabled client-based analytics tool for discovering Singapore property price trends. It aims to provide users with an analytical tool for discovering insights to Singapore property prices through easy-to-understand visual analytics charts. Specifically, it attempts to support the following analysis requirements:

1) To visualise price changes in different areas of Singapore over time;
2) To compare the price range and transaction volume of different areas of Singapore;
3) To identify if the average unit price of a private property project is above or below the average unit price of the planning area and its extent of differences;
4) To enable users to perform statistical analysis on the price distribution of different property types and regions in Singapore.

III. REVIEW AND CRITIC OF PAST WORKS

IV. DATA PREPARATION

Singapore's residential property market is segregated to 2 broad categories: Public HDB Housing and Private Property. The focus of SPIVA is on the private property market i.e. Apartments, Condominiums, Landed Properties. Our data source is REALIS from URA and we are using property transaction records of Year 2018 and 2019.

V. THE APPLICATION SPIVA uses the following libraries to build our app:
For design: shiny, shinydashboard, shinythemes, shinyWidgets, RColorBrewer;
For data wrangling: tidyverse, lubridate;
For plots: geofacet, treemap, ggstatsplot, ggridges.

VI. USER INTERFACE DESIGN

The design of SPIVA dashboard follows the “Visual Information Seeking Mantra: Overview first, zoom and filter, then details-on-demand” by Shneiderman B. [13] while our data visualisations lean towards Edward Tufte’s principal of graphical excellence by maximizing the data ink ratio, within reason [??].

SPIVA consists of four major views namely: (1) transactional overview by planning areas displayed in the form of a Geo-facet, (2) transacted price range comparison of planning areas by planning regions using ridge plots, (3) property price difference of specific projects compared against the mean or median price of the planning area uing Treemap and (4) a price distribution analysis for the statistically-inclined users using violin plot and normality graph (Fig. _). The details of each view will be elaborated below.

SPIVA’s interface (Fig. __) has 2 main sections namely the user selection bar and the visualisation. The selection bar is intuitive to the users and layman terms are used to avoid any confusions or doubts. The selection bar is also interactive and can be hidden to provide a bigger area for the visualisations. We have also allocated a bigger area for the visualisation as it is important that users are able to see every details of the plots we display.

 Fig __: The interface of SPIVA

SPIVA begins with an overview of the Singapore private property market (Fig _). At the top, we will have an information bar providing a quick overview of the Total Units Sold, Average Price and Median Price. Geo-facet is used to provide a high-level trend by planning areas across Singapore. The individual plots are also arranged in the shape of a Singapore map to allow users to identify the planning region that a planning area belongs to.

In all the components, users have the following selections: - Year (2018 or 2019) - Type of Sale (New Sale or Resale) - Property Type (Apartment, Condominium, Executive Condominium, Terraced House, Semi-Detached or Detached House)

For geo-facet overview, users have the following additional selections: - Measure (Average Price, Median Price or Volume) - Axis (Common or Independent) - Grid (Show all or Hide empty)

Fig __: Ridge Plot in SPIVA

The ridge plot is the second visual component of SPIVA. It shows the relationship of total number of transaction against unit price of each planning. For easy visualisation of planning areas that are relatively cheaper to the more expensive ones, the plot is sorted according to the average unit price of each planning area.

For ridge plot, users have the following additional selections: - Region (All, Central Region, East Region, North Region, North East Region or West Region) - View by Planning Area or District Code.

Fig __: Treemap in SPIVA

The treemap is the third visual component of SPIVA. It shows the price differences of different projects compared to the average or median unit price of the planning area. Using treemap as a visualisation is good as users can quickly identify which projects are cheap or expensive by looking at their colour coding. A project that is coloured yellow is cheap while a project that is coloured blue is expensive relative to the average unit price of the planning area they are located. The intensity of the colour shows the extend of difference, therefore, a darker shade means very big difference while a lighter shade means a small difference. Users can zoom in to their particular area of interest by selecting the region and planning area they want to focus on.

For treemap, users have the following additional selections: - Region (All, Central Region, East Region, North Region, North East Region or West Region) - Planning Area (This will be filtered according to users selection of the planning region) - Measure (Average or Median) - Algorithm (PivotSize or Squarified)

VII. USE CASE Our assessment shows that SPIVA possesses several notable advantages in visualizing and analysing Singapore private property prices. With the map view of SPIVA, user can quickly detect usual hotspots where most transactions occur.

Fig. 6 [Show image of SPIVA map view].

Another very important analysis experience discovered in the assessment is the dynamic and interactive features of SPIVA. For example, after detecting a general pattern as shown in Fig. 6, a potential property buyer would like to further explore if [what can he find out from the other visualisations]. He/She can easily accomplish this task by using the interactive [which plot] feature of SPIVA. [With R Shiny App, user can analyse anywhere] This will surely provide the user with more in depth exploration and analysis to the Singapore Private Property data than before he started.

VIII. CONCLUSION

The demonstration of SPIVA’s potential as described in the use case underlined its ability to enable the users to explore and analyse the Singapore private property transaction data to gain valuable insights regarding the price trends, [what else] which in turn could help he/she make more informed choices when looking for the next property purchase.

SPIVA has a lot of potential to be extended and enhanced further. [What are some of the future works to be done]