Lab 9
Name: Teo Yunsheng ID:1001054
Hawker Centre Data
Data Visualization
Is there a relation between area and price?
Below shows the Hawker Centre bid price of different year from Mar 2012 to Jan 2017. Each point on the graph represent the average bid price for a particular floor area stall. The points are colored to indicate the type of stalls.
A simple linear regression line is used to see the relation between area and price. In 2012 to 2015, the data tell us the bids increases as the floor area of the stall increase.
What is the average price per m2 of different stall types?
The bar grap shows that the average bid price for cooked food stall increase steadily from 2013 to 2014. While the average bid price of lockup and market stall did not fluctuate as much. 
How many bids are there for each hawker centre?
From the table, we can conclude that the number of bids for hawker stalls have been increasing steadily during the period of 2013-2016. The graph show the breakdown of the number of bids into type of stalls and year.
| Year | Num of Bids |
|---|---|
| 2013 | 405 |
| 2014 | 548 |
| 2015 | 669 |
| 2016 | 775 |
The density plot below show the number of bids vs the area and they are group by stall type. From the graph, we can see that the market stall have the highest number of bid for approximately 6.5 m2 stalls. While 6.5 m2 and 9 m2 stalls are the 2 main stall type that people bid for.
For lockup stalls, most bidder bid for stall with floor size around 7.5 to 8.5m2.
Spatial Analysis
Is there a spatial distribution with regards to price per m2 of the stall?
Is there a relation between number of bids and average price of the bids?
Number of bids does not correlate well with the average price. The pair have a correlation of -0.11. This is counter intuitive as one would expect that higher number of bid would represent a higher demand which would lead to higher prices.
We can see from the plot below that there are many points with low bids at different price range. This could be due to factors such as stall floor area and its location. Therefore, grouping the data into different sizes, regions or different years would help to check if my assumptions are true.
cor(centres.sp$price,centres.sp$count)## [1] -0.108234
Density Plots
Inserting hawker centres by their latitude and longitude.
centres.sp<- fread("centres_sp.csv",header = T)
centres.sp[is.na(price),price:=0,] #NA to 0
coordinates(centres.sp) <- c('lon', 'lat')
plot(centres.sp)
centres.ppp <- unmark(as.ppp(centres.sp))Read is Singapore shape file and use it as a “window”.
sg <- readOGR(".", "sg-all")## OGR data source with driver: ESRI Shapefile
## Source: ".", layer: "sg-all"
## with 1 features
## It has 13 fieldssg.window <- as.owin(sg)
plot(sg)
Combine the hawker centre position with map of Singapore
centres.ppp <- centres.ppp[sg.window]
plot(centres.ppp)
“Ripley’s K-function” is a method to test if the locations of stationary points are random, clustered or dispersed.
From the plot, the “k isotropic”, “k translation” and “k border” plot are all above the “k poisson” curve. This suggest that the hawker centre location follows a clustered pattern.
If the other lines are very close to the “k poisson curve”, it means that the stall are randomly located. Lastly, if the other lines are below the “k poisson” curve, it means that the location of the points are dispersed.
plot(Kest(centres.ppp))
Heat map indicates that there is high clustering of hawker centre at the southern part of Singapore.
plot(density(centres.ppp),0.02)
contour(density(centres.ppp,0.02))
pop<- as.im(readGDAL("sg-pop.tif"))## sg-pop.tif has GDAL driver GTiff
## and has 37 rows and 58 columnsplot(rhohat(centres.ppp, pop))
plot(rhohat(centres.ppp, pop,weights=centres.sp$price))
plot(pop)
plot(centres.ppp, add=T)
The outliers in areas with high population but low cluster of hawker centres.
The population density heat map plot show that there are two standalone hawker centres at very populated area (pop around 25000). They are Blk 21 Marsiling Lane and Blk 221A Boon Lay Place. Interestingly, Blk 21 Marsiling Lane is located at the north end of Singapore while Blk 221A Boon Lay Place Hawker is located at the far west of Singapore. The ‘dip’ towards the upper end of population density is due to these 2 hawker which are not as clustered compared to other areas of Singapore.
There is high clustering of hawker centre at the southern part of Singapore but the heat map show that the population density of southern part isn’t very high. Why would there be so many hawker centre in a area where the population density isnt that high? One reason is because the population density data is based on where the residents stay. The data does not tell us where people work during the day. The southern area where the hawker centre are clustered is Singapore’s Central Business District (CBD) where many people would travel to work in the day. Despite being in close proximity, these hawker centre are very crowded during lunch hours based on my experience.
In conclusion, the population density heat map plot is only useful to a certain extent. Which bring us back to the point that “Maps do tell lies.” We will need to find alternate ways to study how many people is each hawker centre serving.