Group01 Application

Proposal

Poster

Application

Research Paper

Project Groups

Link to our application

Our app can be accessed here: Hawk_R Stall Rentals

User Guide

Interactive Map Tab	Instructions
	The interactive map tab gives a mapped overview of the location and distribution of rental price trends across different hawker centres in Singapore. Each hawker centre’s location is represented by a blue marker on the Singapore map. When the user clicks on a marker, a pop-up box will display the name of the hawker centre, the number of available cooked food stalls and market stalls. Other information displayed includes the number of bus stops, HDB blocks, HDB carparks and MRT stations within a 350m radius of the hawker centre. [1] An interactive box plot of the average bid price per sqm of all hawker centres in Singapore stall type is embedded at the top of the side tab panel. This allows user to have a visual comparison between the bid price distribution of selected hawker centres and all hawker centres in Singapore. [2] A dynamic time series line graph of average bid price per sqm of selected hawker centre is embedded as the second plot on the panel. User will be able to appreciate changes in bid prices of market and cooked food stall in the selected hawker centre over the years. [3] A dynamic box plot embedded at the bottom of the panel shows the bid price per sqm by stall type in a selected hawker centre. Unlike [1], this box plot is interactive, and updates when a different hawker centre is selected.
Exploratory Tab	Instructions
Sub tab 1: Line graph ]	The exploratory tab contains 4 independent interactive plots to assist user in understanding the distribution of bid price and count across different types of hawker stalls and trade types in Singapore. Sub tab 1 displays an interactive time series line graph of average bid price of market and cooked food stalls across all hawker centres. As user hovers over a point on the line graph, the total number of successful bids for cooked food and market stalls for that corresponding year will be returned as a message at the top right corner of the plot.
Sub tab 2: Scatter plot	The scatterplot of age of hawker against bid price per sqm allows user to have an appreciation of the relationship between recency effect of renovation/ opening on rental prices. Upon hovering across each data point, a pop-up box will return the age of hawker in years and average bid price in the following format (year since last renovation/opening, average bid price per sqm)
Sub tab 3: Bar & Line plot	Sub tab 3 showcases a dual axis bar/line chart where hawker trades are plotted against the number of successful bid counts and average bid price per sqm. Bid count of each trade is represented by the height of the bar in a bar chart and its corresponding average bid price is plotted as a line graph. As user hovers over each bar, the corresponding number of bid count across 2012-2018 will be reflected in a pop-up box. Similarly, the average bid price per sqm of each individual trade will appear in a pop-up box as user hover over the line graph.
Sub tab 4: Treemap	The interactive Treemap incorporates a hover and drill-down function. The number of successful bid counts across 2012-2018 for each trade is represented by the size of the rectangle/ square. The average price per sqm of a trade across 2012-2018 is represented by color of the rectangle/square which increases in shade as rental prices increases. Hover function – As user hover over the plot, all trades under the overarching stall types (market stall vs. cooked food stall) will be highlighted in a darker shade of blue. [1] A pop-up box will reflect the total number of successful bids for cooked food stalls across all hawker centres from 2012-2018. Drill Down function – User can click on either one of the two overarching stall types (market stall vs. cooked food stall) to zoom in to view the proportion of bid counts and bid price per sqm of all trades applicable to that stall type. [2] – At the drilled-down layer, the hover function returns the total number of successful bids for the selected trade across all hawker centres from 2012-2018. [3] – User can click on the < Back icon to return to the overarching view of the treemap
GW REGRESSION TAB	Instructions
	Using the User Input Panel in the Model tab, user can select different combinations of variables to perform GW Regression and interpret the correlations through the colour gradient of the correlation plot generated. Negative correlation is represented by shades of red and positive correlation by shades of blue. Additionally, user can perform Geographically Weighted Regression (GWR) through selection of regression type, kernel functions, independent and dependent variables. A model comparison table shows the fit statistics of both the global linear model (LM) and GW Regression model. Once the regression is run, user can see which variables that are significant at 90%, 95%, 99% and 99.9% level. The GWR Plot tab gives a localised coefficient and p-value for each hawker centre, and we can identify clusters of hawker centres that show similar characteristics. User can select the variable to explore and see how that impacts each hawker centre. There are 2 sub tabs for user to toggle between - Models and GWR Plots.
Sub tab 1: Models	COMPONENTS OF USER INPUTS PANEL User may choose the parameters to input for the GW Model as follows, Type of stall – Cooked Food Stalls, Market Stalls Select Kernel – Gaussian, Exponential, Box-Car, Bo-Squared, Tri-Squared Dependent Variable – Average Price psm, Median Price psm Select Independent Variables – All, None Regression Type – Basic GWR, Robust GWR – Filtered, Robust GWR - Downweight [1] Significance level – After performing the GW Regression, user can see which variables are significant at 90%, 95%, 99% and 99.9% level. [2] Model Comparison – User can compare how Geographically Weighted Regression (GW) model perform when benchmarked against the Global Linear Regression Model (LM) using the various metrics. [3] Global GW Regression Model Coefficients & p-value – User can compare the significance of each individual inputs in the GW model
Sub tab 2: GWR Plot	[1] – User may toggle between 21 input variables to view the coefficient and p-value bubble plots for each variable. [2] – User can download the results of the GW Regression with the following information – name of hawker centre, longitude, latitude, intercept, coefficient, actual dependent variable (y), predicted dependent variable (yhat), residual value, standard error t-value and p-value. [3] – User can select to view up to 10, 25, 50, 100 rows of data at one time. [4] – User may sort the results table according to the coefficient, p-value, actual dependent variable, predicted dependent variable and R² value.