ISSS608 2016-17 T1 Assign1 Kee Bei Jia

From Visual Analytics and Applications
Jump to navigation Jump to search

Abstract


Using Tableau and the pubic data set on Resale Flat Prices, the trend, seasonality and market shares can be easily observed using visual analytics. Comparisons have also been done between the first half of the year of 2015 & 2016.

From the info graphic below, it can be seen that Resale prices are depended on location with Locations like Central Area, Bukit Timah, Bishan, and to my surprise Bukit Merah, while the resale price is high actual transactions are low for those areas, showing that the location is valuable. And most transactions occur with 3-5 Room Flats (28%,40%,23% respectively), and Model of Improved(29%), Model A (26%), and New Generation (17%).

There is a slight upward trend in Resale prices (judging from the median resale price in the Box and Whiskers plot), as well as seasonality in Transactions across the months.The month with the lowest transaction is February and highest transaction April and May.


Problem & Motivation


Singapore is an Island country with limited land area and dense population. and the population is expected to increase in the future. As such housing is a problem, to as when, where and what to buy and its price as there are speculations about the rising cost of housing in Singapore.

The main variables I would be looking at are:

  1. Number of Transactions
  2. Resale Prices
  3. Town
  4. FLoor Area Sqm

With Visualization techniques, the trends and current state of housing prices can be easily interpreted and understood.

Approach


To represent shares of mousing in the market, I went with number of transaction and against dimensions like Flat Type and Flat Model. The bar chart, circle view and Pareto chart all shows the different dimension of flats in the market.

For distribution of sales price, Histogram and Box & Whiskers plot are able to portray the distribution.

For comparison, I used line charts to portray Trends and Box & whiskers plot to show the variance in prices. The comparisons are broken down by Flat Type, Town and Dates Transacted. Transactions broken down by Flat Type/Town and plotted against Month and year, showed that the trend or seasonal patterns present are consistent across years 2015 & 2016.

The Box & whiskers plot shows the variance of resale price across different locations. The high and low outliers are consistent across years 2015 & 2016.


Limitations:

  • Tableau have a limit of 20 color palettes for their legend. As there are more towns than colors, the colors repeat on the charts, and this is not ideal given that we are using it as a non-interactive info-graphics.
  • Tableau is missing the function to allow users to rotate axis labels (ones under the marks) by 45 degrees, there is only horizontal and vertical, as such some of the words might be harder to read.

Tools Utilized


Tableau was used.

Data preparations: There are some transformation required before the chart building starts:

  1. Filters are used to filter out only the 2015 & 2016 transaction as required by the charts.
  2. Conversion of Date field: The Date field in the data set was in the format YYYY-MM, as in “2012-03” for March 2012 as type string. To convert to a date format, the following formula was used on the Month field: DATEPARSE("yyyy-MM",[Month] )
  3. Calculating the price per square meter of resale prices: The prices in the data set was the transaction price.

As such, for a fairer comparison, Price per square meter was calculated using the formula: [Resale Price]/[Floor Area Sqm]


Preparation of info-graphics: To create a histogram, a bins dimension was created from the Price per square meter measure, bin size was dependent on a parameter bin size which could be altered in Tableau dashboard. Filters could also be added in for further drill down if required. (They were left our as it was an info graphic.)

The info graphic was also prepared in Tableau Dashboard tab, and exported from there as PNG file.

Resale Flat Info-graphics

Results


General observations


What are the shares of the resale public housing supply in 2015 (at least three observation):

1) Among the resale flats, 3 flat type dominated the market: 4 room flats (40%), followed by 3 room flat (28%) and 5 room flats(23%) of all sales in 2015.

2) Among those sales, the most transacted models are Improved for 5 room flat(16.7%), Model A for 4 room flat (22.6%) and New Generation for 3 room flat (11.6%)

3) The market is predominantly by Improved flats (29%), Model A (26%) and New generation (17%).


What are the distribution of the resale public housing prices in 2015 (at least three observation):

1) Even though resale flats with more square meters generally cost more, the mean and median price per sqm distribution for 5 room flats are lower than that of 4 rooms ,and 4 rooms less than that of 3 rooms distribution. (Judging from the histogram and box & whiskers plot) With the exception of Executive and Multi-Generation Flat Type.

2) The resale price of 1 room flats and Multi-generation flats doesn't vary much, but this could be due to lack of transaction in those category. For example, there is only 1 Transaction of Multi-generation flat.

3)But from 2 to 5 room flats and Executive, the resale price vary also in the box and whiskers plot as they were affected by other factors such as location of town and story range. The variance mainly stem from locations such as Central Area and few transactions from Seng Kang, Bedok and Kallang/Whampoa.


With reference to the findings, compare the patterns of the first-half of 2016 with the patterns of 2015:

1)The total transaction records showed similar trend in both 2015 and 2016, suggesting that there could be seasonal factor with a slight upwards trend in prices (judging from the median resale price in the Box and Whiskers plot). With a spike in transaction in April and May, and low transactions in February. The effects are mainly felt by the 3 to 5 room flats.

2) Despite the high prices per square meter for housing in the Central Area, Bishan, Bukit Merah and Bukit Timah pert of town, the number of transactions for housing in those location were low. Perhaps one of the reason the prices are high, for both year 2015 and 2016.

3) The prices for flats in areas such as Woodlands, Jurong West are consistently in the lower range, but they have a lot of transactions in both 2015 and 2016.