Anda di halaman 1dari 47

2013

GISC 9303-Spatial Analysis D4b Geostatistical Analysis of Housing Sale Prices in St. Catharines Neighborhoods

Prepared For: Dr. Ian Smith Prepared By: Nobuhle Matanga, Lisa Atkinson and Stephanie Korhonen

March 22, 2013

March 22, 2013 File: GISC93084b

Mr. Ian D. Smith M.Sc., OLS, OLIP, EP Post-Graduate Professor of Environmental Sciences and GIS Niagara-on-the-lake Campus, Niagara College Room E313 135 Taylor Road, S.S #4 On, L0S 1J0 Dear Mr. Smith: Re: Submission of GISC93084b Please accept this letter as the formal submission of Deliverable 4b Geostatical Report for GISC9308 Spatial Analysis. This geostatistical report outlines all intentions for undertaking the statistical investigation on the Average Housing Sale Prices in St. Catharine Neighborhoods. The attached report beings by highlights the study area, objectives, goals, methodologies and preliminary statistical assessment. This report concludes with a comparison of the IDW and Kriging results created from the obtained dataset. Supplemented material include: a summary of collected raw data, statistical calculations, and graphing of the dataset. Overall it was determined that the IDW method resulted in a better prediction surface. Please do not hesitate to contact us for any additional information at 1-289-241-7627. Thank you for your time and attention. Sincerely,

Nobuhle Matanga, B.Sc. GIS-GM Graduate Candidate GIS Tek. N.M. / Enclosures: 1) [Geostatical Collection Report], 2) [Statistical Calculations and Graphing of Data Set]. Cc. Lisa Atkinson, BA Stephanie Korhonen, BA
135 Taylor Road, S.S #4 Niagara-on-the-lake, ON Tel: 1-289-241-7627 Email: nobuhlematanga@gmail.com

March 22, 2013

Executive Summary
This geostatistical report begins by providing an in-depth summary of the study area, objectives, goals methodologies and preliminary statistical assessment of the Housing Sale Prices in St. Catharine Neighborhoods, for the obtained dataset. The geographic data is defined as

UTM Easting and Northing coordinates; whereas the z value is presented as housing cost. Formal maps, displaying the housing locations and the study area extent are located within Appendix A, and a full glossary of terms is located in Appendix B. The purpose of this investigation is to determine the feasibility of this dataset for future geostatistical studies. The St. Catharines housing price dataset can be summarized as follows; there are a total of 138 observations used in this study, the mean price is $367,270, the median price is $259,950, more importantly there is a kurtosis of 22.14, skewness of 3.9814 and a standard deviation of 2188.1. As a result of the positive skew in this dataset, a log transformation is required before spatial importation can be conducted. The second half of this report discusses the prediction surfaces created using both the kriging and IDW techniques. Although both surfaces provide adequate representations of wealth zones in the St. Catharines area, the kriging results are more skewed due to outliers. Therefore it was concluded that the IDW provides a more accurate classification of poverty and affluence in St. Catharines area. As a result the information from the IDW results will essentially allow contractors to maximize profit and minimize cost.

Geo Tek | Geostatistics Report

March 22, 2013

Table of Contents Executive Summary ..................................................................................................................... i 1.0 Project Understanding ............................................................................................................ 1 1.1 Study Area ............................................................................................................................. 1 1.2 Project Goal ........................................................................................................................... 2 1.3 Objectives and Benefits of this Project this Project .............................................................. 2 2.0 Summary of Methodology ..................................................................................................... 2 2.1 Data Collection ...................................................................................................................... 2 2.2 Determining Sample Size ...................................................................................................... 2 2.3 Formatting and Displaying Data in ArcGIS .......................................................................... 3 2.3.1 Creation of a file Geodatabase ................................................................................... 3 2.3.2 Creating Metadata for file Geodatabase .................................................................... 3 2.3.3 Importing X, Y data into ArcGIS .............................................................................. 3 2.3.4 Projecting Data........................................................................................................... 3 2.3.4 Georeferencing Neighborhood Boundaries Data ....................................................... 3 2.4 Geostatistical Analysis of Data in ArcGIS ............................................................................ 4 2.4.1 Summary statistics ..................................................................................................... 4 2.4.2 Histogram ................................................................................................................... 6 2.4.3 Normal QQ Plots........................................................................................................ 8 2.5 Kriging Interpolation ........................................................................................................... 12 2.6 Inverse Distance Weighted (IDW) Interpolation ................................................................ 17 3.0 Results and Discussion ...................................................................................................... 19 3.1 Kriging vs. Inverse Distance Weighted (IDW) ................................................................... 19 3.1.1 Similarities ............................................................................................................... 19 3.1.2 Differences ............................................................................................................... 20 4.0 Conclusions ......................................................................................................................... 25 5.0 References ........................................................................................................................... 26 APPENDIX A (Formal Maps) .......................................................................................................... APPENDIX B (Glossary of Terms and Parameters) ........................................................................ APPENDIX C (Raw Data)................................................................................................................

Geo Tek | Geostatistical Report

ii

March 22, 2013

List of Figures Figure 1 : Formal Map of Study Area ............................................................................................. 1 Figure 2 Summary Statistics Tool in ArcGIS, Image Source ArcGIS ........................................... 4 Figure 3: Summary statistics for Easting Observations, Image Source ArcGIS ............................. 5 Figure 4: Summary statistics for House Prices, Image Source ArcGIS .......................................... 5 Figure 5: Summary statistics for Northing Observations, Image Source ArcGIS .......................... 5 Figure 6 Northing Observations Histogram, Image Source ArcGIS.............................................. 6 Figure 7: House Prices Histogram, Image Source ArcGIS ............................................................. 7 Figure 8 Easting Observations Histogram, Image Source ArcGIS ................................................. 7 Figure 9 House Price Normal QQPlot, Image Source ArcGIS ....................................................... 8 Figure 11: Northing Observation QQPlot, Image Source ArcGIS .................................................. 9 Figure 10: Easting Observations QQPlot, Image Source ArcGIS................................................... 9 Figure 12 Normal QQ plot of Housing Prices with Log Transformation, Image Source ArcGIS10 Figure 13 Histogram of House Prices with Log Transformation, Image Source ArcGIS ............ 11 Figure 14 : Study Semivariogram, Image Source ArcGIS ............................................................ 12 Figure 15: Kriging Parameters Used In Study ............................................................................. 14 Figure 16: IDW Parameters Used In Study .................................................................................. 17 Figure 17: House Price Zone Classification, Image Source ArcGIS ........................................... 19 Figure 18: Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Quadrant 1..................................................................................................................................................... 21 Figure 19: Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Quadrant 2..................................................................................................................................................... 22 Figure 20: Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Quadrant 3..................................................................................................................................................... 23 Figure 21: Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Quadrant 4..................................................................................................................................................... 24 Figure 22: Example of Spatial Interpolation, Image Source: Niagara College ............................. 2 Figure 23: Kriging Calculation, Image Source: Niagara College .................................................. 3 Figure 24: IDW Calculation, Image Source: Niagara College ...................................................... 4

List of Tables Table 1 : Cross Validation Assessment of Kriging Results .......................................................... 15 Table 2:Cross Validation Assessment of IDW Results ............................................................... 18

Geo Tek | Geostatistical Report

iii

March 22, 2013

1.0 Project Understanding


1.1 Study Area
The scope of this study is the St Catharines region. The study area is been divided into 54 subsections, encompassing 28 St Catharine's neighborhoods. A total of 138 observations are to be assessed. The study area is defined by Figure 1, below:

Figure 1 : Formal Map of Study Area

March 22, 2013

1.2 Project Goal


The goal of this spatial statistical assessment is to determine areas of poverty and affluence in St Catharine's, using sample residential housing sale prices.

1.3 Objectives and Benefits of this Project this Project


This project will allow for a practical assessment of future building projects, within a specified neighborhood, in order to maximize profit, and minimize cost.

2.0 Summary of Methodology


2.1 Data Collection
The housing cost and address data is collected from the Relators Canada Incorporated website. This website is owned by the Canadian Real Estate Association and the National Association of Realtors (Realtors.ca, 2013). The data available on the website is provided by realtors from across Canada, and is updated hourly (Realtors.ca, 2013). A Multiresolution Seamless Image Database (MrSID) file, of the Niagara region and corresponding municipalities boundaries, is provided by Niagara College. Subsequently, UTM NAD 83 Zone 17N Easting and Northing coordinates are obtained via Google Earth. Google Earth is a real world representation of superimposed images obtained from satellite imagery, aerial photography, and GIS 3D globe. This platform is available at no cost to public users.

2.2 Determining Sample Size


All the sample data was collected on January 25, 2013. This dataset includes residential houses that are for sale. On January 25, 2013 there were a total of 360 houses for sale in St Catharines. However, in order to avoid large generalizations and minimize inaccuracies, the St. Catharines region was divided into 54 equal area subsections. From these subsections

maximum, median and minimum values were obtained, therefore reducing the sample size to 138 observations.

Geo Tek | Geostatistical Report

March 22, 2013

2.3 Formatting and Displaying Data in ArcGIS


2.3.1 Creation of a file Geodatabase The file geodatabase associated with this deliverable is created using ArcCatalog and is set as the default geodatabase within the map document properties. This ensures a common projection for all assessment products, and ensures products are exported to the correct geodatabase, and finally, reduces project costs associated with data transfer. 2.3.2 Creating Metadata for file Geodatabase Metadata (tags, summary, description, credits and use limitations) are created for the file geodatabase using ArcCatalog. Metadata is essential for data management purposes; it provides the user with information such as data source and data function. 2.3.3 Importing X, Y data into ArcGIS An excel table composed of longitude, latitude and price values is transformed to a shapefile via ArcGIS. The Easting coordinates are assigned as X values, the Northing

coordinates are assigned as Y values, and the price observations are assigned as Z values within the attribute table. 2.3.4 Projecting Data All of the imported data was reprojected into UTM Zone 17N, NAD 1983, this is the desired format for all the data used in this study. 2.3.4 Georeferencing Neighborhood Boundaries Data Georeferencing is the process of assigning raster data sets to a map coordinate, positional reference, system (Smith, 2013). The purpose of this is to rectify data. Thus, a jpeg image of St Catharine's neighborhood boundaries is correlated to the MrSID file of the Niagara Region, via the Georeferencing tool in ArcGIS. This data management tool is used to allocate three control point pairs to each image to warp the jpeg images to the MrSID referenced data. Control points are geographic references, easily identifiable upon both the non-referenced image and the referenced image. When georeferencing raster data, at least three, well distributed control points must be established to ensure precise image warp effects.

Geo Tek | Geostatistical Report

March 22, 2013

2.4 Geostatistical Analysis of Data in ArcGIS


2.4.1 Summary statistics Summary statistics (Count, Min, Max, Mean, Std. Dev., Range, Frequency and Sum) for the St. Catharine's housing prices, Easting and Northing coordinates, are investigated utilizing the Geostatistical Analyst tool bar in ArcGIS. The summary statistics for the housing prices, Northing and Easting observations are displayed in Figures 2 to 5.

Figure 2 Summary Statistics Tool in ArcGIS, Image Source ArcGIS

Geo Tek | Geostatistical Report

March 22, 2013

Figure 4: Summary statistics for House Prices, Image Source ArcGIS

Figure 3: Summary statistics for Easting Observations, Image Source ArcGIS

Figure 5: Summary statistics for Northing Observations, Image Source ArcGIS

Geo Tek | Geostatistical Report

March 22, 2013

2.4.2 Histogram Histograms for the St. Catharines housing prices, Easting and Northing coordinates are created using the Geostatistical Analyst tool bar in ArcGIS. The number of bins used for our histograms (12) is determined by rounding the square root of the sample size (138). This was rounded because the square root of 138 is 11.74 and a decimal number cannot be used for the number of bins. The histograms are used to assess the frequency distribution of within the dataset. the values

The presented histograms, display statistical analyses provided by the

Geostatistical Analyst tool such inclusing, skewness, kurtosis, median, 1-st Quartile and 3-rd Quartile. The Northing observations are normally distributed while, the Easting values have a negative skew and the price observations have a positive skew (Figures 6-8).

Figure 6 Northing Observations Histogram, Image Source ArcGIS

Geo Tek | Geostatistical Report

March 22, 2013

Figure 8 Easting Observations Histogram, Image Source ArcGIS

Figure 7: House Prices Histogram, Image Source ArcGIS

Geo Tek | Geostatistical Report

March 22, 2013

2.4.3 Normal QQ Plots Normal QQ plots for the St. Catherines house prices, Easting and Northing coordinates, are created utilizing the Geostatistical Analyst tool bar in ArcGIS. The standard normal distribution quantile values are represented by the x-axis of a Normal QQ Plot, and the dataset quantile values are represented by the y- axis. If data values are normally distributed they will be plotted in proximity to a computed reference line. The Easting and Northing data (Figures 10-11) sets are normally distributed, in contention with the pattern observed for the housing price data set (Figure. 9).

Figure 9 House Price Normal QQPlot, Image Source ArcGIS

Geo Tek | Geostatistical Report

March 22, 2013

Figure 11: Easting Observations QQPlot, Image Source ArcGIS

Figure 10: Northing Observation QQPlot, Image Source ArcGIS

Geo Tek | Geostatistical Report

March 22, 2013

As previously stated, based on the QQ plots of the Easting and Northing data values, a normal distribution exists, while the house prices are not normally distributed. The house prices appear to have a positive skew. As a result a log transformation is applied to the price data to normalize the data. The results of the log transformation are displayed in the normal QQPlot (Figure12) and the histogram (Figure 13).

Figure 12 Normal QQ plot of Housing Prices with Log Transformation, Image Source ArcGIS

Geo Tek | Geostatistical Report

10

March 22, 2013

Figure 13 Histogram of House Prices with Log Transformation, Image Source ArcGIS

Geo Tek | Geostatistical Report

11

March 22, 2013

2.5 Kriging Interpolation


Before conducting the kriging technique a variography (structural analysis), must be conducted on the dataset. This variography is conducted by created an empirical semivariogram. An empirical semivariogram plots the relationship between distance and average semivariance. Overall in this studys empirical semivariogram proves that as distance increases the observations become more variable. The semivariogram in this study follows the Gaussian (hyperbolic) model, more specifically there is increase in semivariance with distance, shown in figure 14 below:

Figure 14 : Study Semivariogram, Image Source ArcGIS

Based on this graph it can be inferred that the housing price dataset for the city of St. Catharines is very continuous at close distances. Additionally it be predicted that the prediction surface created from this model would be very smooth, there would be a lot of gradual shifts between price classes.

Geo Tek | Geostatistical Report

12

March 22, 2013

Additionally the pre-kriging structural analysis determine that the housing price dataset for St. Catharines is anisotropic (directional). Housing price is depends on direction and distance. The reason for this is that houses that are closer together generally cost the same. Moreover housing cost is also depended on geographic location, for example housing cost tends to increase going towards waterfront properties and decreases the inner city, and this is illustrated in both the IDW and Kriging surface results

In order to interpolate the housing cost across a surface, ordinary kriging is employed to display the housing cost variance. For this investigation, ordinary kriging is utilized, so that the constant mean is assumed as unknown, as is a common best practice of geostatistical analysis (Smith, 2013). The Kriging results depend on the semivariogram model. This technique classifies data using the semivariogram and relative distance. Additionally, kriging settings are enforced, via the interactive kriging tool window. These settings are summarized by Figure 15, shown below:

Geo Tek | Geostatistical Report

13

March 22, 2013

Figure 15: Kriging Parameters Used In Study

Geo Tek | Geostatistical Report

14

March 22, 2013

The input parameters above, summarized as Figure 14 , creates a data depiction utilizing semivariogram; comparing data points, in terms of local neighbours, and similarity of housing cost, based on distance to neighboring data points (Smith, 2013). The cross validation graphs are important tools to understand the predictability of a data set, or investigation. For this discussion, the cross variance outputs, and interpretations, are summarized by Table 1:
Table 1 : Cross Validation Assessment of Kriging Results

Cross Validation Prediction

Depiction

Interpretation This graph displays the prediction of a normal distribution in grey. However, the blue line represents the predictability of the housing cost data presented in this study. While not a perfect model of prediction, this data is not displayed as random either. Within the confines of the investigation, the prediction trend is logical for housing costs. Due to the presence of a selection of extremely high housing costs, the majority of data points appear clustered. This in turn, affects the prediction model for the entire surface being examined. However, these points are not outliers, as they represent lake front properties. The standard error is also a measurement of the success of a prediction model, such as kriging. Again, the data points appear clustered due to extreme housing costs for lake front properties. Therefore, the ability to predict housing costs accurately, across the entire study area is affected, and may contains errors.
Geo Tek | Geostatistical Report 15

Error

Standard Error

March 22, 2013

Normal QQ Plot

The predicted surface for the kriging result, based on the localization of data points, among the normal distribution line, will result in an accurate, and normal, prediction surface.

Geo Tek | Geostatistical Report

16

March 22, 2013

2.6 Inverse Distance Weighted (IDW) Interpolation


As a supplementary check, on the predictability of the surface, a second method of creating a predicability surface model is completed. This model will display the housing cost variance over a surface. Where Kirging appears to give a more smooth transition of data values avoiding the bulls eye effect and gives a standard error, IDW is more simplified, requiring less user input parameters to produce a result (Smith, 2013). The IDW is completed for the cost variable of this study. IDW settings are enforced, via the interactive geostatistical wizard tool window. These settings are summarized by Figure 16, shown below:

Figure 16: IDW Parameters Used In Study

Geo Tek | Geostatistical Report

17

March 22, 2013

The input parameters above, summarized by Figure 16, creates a data depiction, comparing data points, in terms of local neighbours, and similarity of housing cost, based on distance to neighbouring data points. The cross validation graphs, for the IDW results, are important tools to understand the predictability of a data set, or investigation. For this discussion, the cross variance outputs, and interpretations, are summarized by Table 2:
Table 2: Cross Validation Assessment of IDW Results

Cross Validation
Predicted

Depiction

Description
This graph displays the prediction of a normal distribution in grey. However, the blue line represents the predictability of the housing cost data presented in this study. While not a perfect model of prediction, this data is not displayed as random either. Within the confines of the investigation, the prediction trend is logical for housing costs. Due to the presence of a selection of extremely high housing costs, the majority of data points appear clustered. This in turn, affects the prediction model for the entire surface being examined. However, these points are not outliers, as they represent lake front properties. Further, as compared to the Kriging results, there is less of a data distribution stretch, due to the bull-eye nature, of this method.

Error

Geo Tek | Geostatistical Report

18

March 22, 2013

3.0

Results and Discussion

3.1 Kriging vs. Inverse Distance Weighted (IDW)

3.1.1 Similarities The kriging and IDW methods produce surface covers that categorize St. Catharines into different zones based on housing price. These zones are classified according to the parameters outlined in Figure 17, below:

Figure 17: House Price Zone Classification, Image Source ArcGIS

Both the Kriging and the IDW surfaces show a decrease in housing prices towards the core of the city, downtown St. Catharines, as well Riverview. Additionally, both surfaces display an increase in housing prices towards the greater Louth area, Western Hill, Burleigh Hill and lake front properties, as shown in Appendix A. Both surfaces have excellent coverage and no data daps, or holes are present in the data.

Geo Tek | Geostatistical Report

19

March 22, 2013

3.1.2 Differences In the Kriging surface there is a more pronounced presence of extreme housing prices, both high and low. In comparison to the IDW results, a more gradual transition between areas of high housing costs and low housing costs is present. Due to the generalization of the IDW results there is a larger error for predicting housing cost, as opposed to the Kriging results, which show a greater amount of localized detail. For a better assessment of the differences the larger study area, is divided into quadrants.

Geo Tek | Geostatistical Report

20

March 22, 2013

In quadrant 1, the differences between the two interpolation methods are most apparent in the Martindale neighborhood, particularly the area surrounding the $2,150,000.00 Martindale home. In the IDW results there is a gradual increase in price towards the center of the neighborhood where a bulls-eye effect occurs. In comparison, the kriging results predict that the majority of the houses in Martindale will be expensive and not just those within in the direct vicinity of the $2,150,000.00 dollar Martindale home. However based on the dataset, delineating housing costs around the $400,000 cost, the IDW results appear to be more accurate for this quadrant. These comparisons are shown by Figure 18, below:

Figure 18: Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Quadrant 1

Geo Tek | Geostatistical Report

21

March 22, 2013

In quadrant 2 both the kriging and the IDW results are relatively similar. The only notable difference is that the IDW results tend to account for average housing prices, whereas the kriging result are highly influenced by the more extreme values. However, as indicated by the original dataset figures, houses in and around the North End are relevantly cheap in comparison, with the exception of waterfront property. In this particular quadrant the kriging results appear to be more correct. These comparisons are displayed by Figure 19, below:

Figure 19: Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Quadrant 2

Geo Tek | Geostatistical Report

22

March 22, 2013

In quadrant 3,

there are differences in the prediction for the Louth neighborhood

(located in the southwest region of the maps below). In the kriging results, there is a smaller area for houses with extremely high prices, whereas this area is expanded in the IDW results. Moreover, there is larger area for moderately high housing costs, in comparison this area is a lot smaller in the IDW results. In this quadrant it appears that the outlier, the $1,995,000 house, greatly skews the IDW results. Therefore, in this quadrant the kriging results are more accurate. These comparisons are displayed by Figure 20, below:

Figure 20: Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Quadrant 3

Geo Tek | Geostatistical Report

23

March 22, 2013

In quadrant 4, the only major difference between the kriging results and the IDW is the increase of designation of areas as extremely low cost housing in the kriging results. Overall both of these results are fair and give an accurate prediction of housing costs of this area. These comparisons are displayed by Figure 21, below:

Figure 21: Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Quadrant 4

Overall the IDW results give a more accurate prediction of the housing cost in the study area.

Geo Tek | Geostatistical Report

24

March 22, 2013

4.0

Conclusions
To conclude, an in-depth summary of the study area, objectives, goals, methodologies

and preliminary statistical assessment of the House Sale Prices in St. Catharine Neighbors study dataset was outlined in the beginning of this report. Based on the results of the statistical assessment, the dataset was declared suitable for a geostatistical analysis. A geostatistical

analysis, utilizing both kriging and IDW interpolation methods, was conducted on this study area. Both of these methods produced prediction surfaces that divided St. Catharines into different zones based on housing price. A comparison of these two surfaces reveals that the

IDW results is less skewed towards outliers and is therefore the more representative surface. In turn the IDW results correctly determine areas of poverty and affluence in St. Catharines which will influence of the location of future building project in this city. The IDW surface will essentially allow contractors to maximize profit and minimize cost.

Geo Tek | Geostatistical Report

25

March 22, 2013

5.0

References
Smith, Ian. Week 1- Introduction to Stats. GISC9308-Spatial Analysis. Niagara College. PDF. Smith, Ian. Week 2- Multivariate Statistics. GISC9308-Spatial Analysis. Niagara College. PDF. Smith, Ian. Week 3- Sampling. GISC9308-Spatial Analysis. Niagara College. PDF. Smith, Ian. Week 4- Introduction to Spatial Analyst. GISC9308-Spatial Analysis. Niagara College. PDF. Smith, Ian. Week 7- Regression and Interpolation. GISC9308-Spatial Analysis. Niagara College. PDF. Smith, Ian. Week 8- Geostatistical Analyst. GISC9308-Spatial Analysis. Niagara College. PDF.

5.1. Lectures

5.2 Software ArcGIS (2008) ArcGIS Desktop Education Edition (Version 10). Computer program. Available at http://www.esri.com/products 5,Sept, 2012

5.3 Terms of Reference "Assignment4- Geostatistical Analysis of Student Collected Spatial Data." GISC9308-Spatial Analysis. Niagara College. Web.

5.4 Textbook Ormsby , Napoleon , Burke , Carolyn Groessl and, Laura Bowden. Getting to Know ArcGIS Desktop; for ArcGIS 10. Redlands: Esri press. 2010. Print.

5.5 Websites "ArcGIS Resources." ArcGIS. Esri, n.d. Web. 20 Jan 2013.
Geo Tek | Geostatistical Report 26

March 22, 2013

<http://resources.arcgis.com>. "Realtors.ca." The Canadian Real Estate Association. Web. 20 Jan 2013. <http://www.realtor.ca/>.

Geo Tek | Geostatistical Report

27

March 22, 2013

APPENDIX A (Formal Maps)

March 22, 2013

A.1 Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Study Area

March 22, 2013

A.2 Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Kriging Results

Geo Tek | Geostatistical Report

March 22, 2013

A.3 Real Estate Sale Prediction, St. Catharines, Ontario and Surrounding Area: Inverse Distance Weighting Results

Geo Tek | Geostatistical Report

March 22, 2013

APPENDIX B (Glossary of Terms)

March 22, 2013

B.1 Spatial Interpolation


Spatial Interpolation is the process of using known values (kernels) to mathematically compute unknown values. In the ArcGIS geostatistical extension this can be used to create prediction surfaces incorporating known and predicted z-values. The example, shown in figure 22, below shows use the z- values of 4 data points are interpolated and used to compute 12 predicted z-values.

Figure 22: Example of Spatial Interpolation, Image Source: Niagara College

This study will seek to use 138 kernels to compute predicted z-values for the entire city of St. Catharines. Additionally, this study will compare the results of two different spatial interpolation methods available in ArcGIS geostatistical extension, the kriging technique and the IDW technique. Before conducting either of these spatial interpolation methods, a data analysis must be conducted. Based on the histogram, QQ plot of the St. Catharines, housing price dataset it was determined that the house prices appear to have a positive skew. As a result a log transformation was undertaken on the price data to normalize the data. It was also noted that there are directional influences on the data and therefore the neighborhoods were not divided in to sectors. This information will be incorporated in the interpolation of this dataset.

Geo Tek | Geostatistical Report

March 22, 2013

B.2 Kriging Technique


The kriging technique creates surfaces were by predicted z-values are based statistical relationships between kernels; this is referred to as the autocorrelation method. This method is based on the following equation. Before conducting the kriging technique a variography (structural analysis), must be conducted on the dataset. This variography is conducted by created an empirical semivariogram. An empirical semivariogram plots the relationship between distance and average semivariance. Overall in this studys empirical semivariogram proves that as distance increases the observations become more variable. The empirical semivariogram is this study follows a Gaussian trend. The equation for this method is outline in figure 23.

Figure 23: Kriging Calculation, Image Source: Niagara College

In ArcGIS there are two different kriging methods available, the ordinary and the universal. The ordinary method is to be used on data that is variable, whereas the universal method is used on data that follows a trend. The ordinary kriging method is used is this study in order not to bias results, more specifically this technique was conducted under the assumption that the dataset is trendless. The kriging method is more applicable for datasets that are highly variable.

Geo Tek | Geostatistical Report

March 22, 2013

B.3 Inverse Distance Weighted (IDW) Technique


The IDW technique creates surfaces were by predicted z-values are directly based on surrounding kernels; this is referred to as the deterministic method. This method is based on the following equation. The equation for this method is outline in figure 24.

Figure 24: IDW Calculation, Image Source: Niagara College

Additionally, the IDW method is more applicable for datasets were distance greatly effects influence.

Geo Tek | Geostatistical Report

March 22, 2013

APPENDIX C (Raw Data)

March 22, 2013

Grid Number
1 1 1 1 2 2 2 3 3 3 4 4 4 4 5 5 5 5 6 6 6 7 7

Neighbourhood
North End North End North End North End North End North End North End Port Weller Port Weller Port Weller Port Dalhousie Port Dalhousie Port Dalhousie Port Dalhousie Michigan Beach Michigan Beach Michigan Beach Port Dalhousie North End North End North End North End North End

Price
$2,799,000.00 $339,900.00 $364,900.00 $689,900.00 $102,500.00 $189,900.00 $269,900.00 $189,500.00 $329,900.00 $569,000.00 $179,000.00 $299,500.00 $260,000.00 $294,000.00 $1,599,900.00 $559,900.00 $589,900.00 $274,900.00 $799,000.00 $209,000.00 $299,000.00 $189,900.00 $244,900.00

Address
15 Lantana Circle , St Catharines 5 lakebreeze Crescent , St Catharines 1 Warrington Place, St Catharines 39 Royal York Road, St Catharines 11 Grandview Drive, St Catharines 78 Melody Trail, St Catharines 103 Arthur Street, St Catharines 9 Shoreham Street, St Catharines 6 Moes Crescent, St Catharines 4 Yonge Street, St Catharines 65 Main Street, St Catharines 10 Ann Street, St Catharines 94 Dalhousie Avenue , St Catharines 99 DalhousieAvenue , St Catharines 14 Shore Boulevard, St Catharines 5 Xavier Court, St Catharines 3 Cricket Hallow Road, St Catharines 27 Simpson Road, St Catharines 164 A Lakeshore Road, St Catharines 25 Murray Street, St Catharines 13 Costen Boulevard, St Catharines 584 Bunting Road, St Catharines 20 Pearce Avenue, St Catharines

Easting
643553.52 643761.79 643433.92 643188.45 644447.55 644698.48 644376.37 645284.67 645564.06 645093.33 640511.41 640464.39 640327.44 640261.58 641820.34 642478.20 642107.59 642143.90 643081.77 643747.83 643807.56 645313.51 644903.32

Northing
4786445.99 4786315.68 4785962.26 4786135.16 4787008.56 4786792.85 4786942.22 4787223.03 4787138.55 4787070.34 4784479.55 4784434.50 4784526.88 4784467.25 4785262.42 4785079.13 4785154.62 4784860.87 4785220.63 4785657.60 4785198.63 4785289.58 4785637.49

March 22, 2013

7 7 9 10 10 10 10 11 11 11 12 12 12 13 13 13 14 14 14 14 15 15 15 15 17

North End North End Louth Martindale Martindale Martindale Port Dalhousie Port Dalhousie Port Dalhousie Port Dalhousie North End North End North End North End North End North End North End North End North End North End North End North End North End North End Louth

$269,900.00 $229,900.00 $750,000.00 $2,150,000.00 $448,500.00 $439,900.00 $229,000.00 $869,000.00 $244,900.00 $399,900.00 $315,000.00 $189,900.00 $219,900.00 $349,000.00 $179,900.00 $249,900.00 $199,900.00 $339,000.00 $229,900.00 $225,000.00 $167,500.00 $219,900.00 $315,900.00 $209,400.00 $519,900.00

23 The Cedars Street, St Catharines 7 Glencairn Drive, St Catharines 1214 Lakeshore Road, St Catharines 1010 Lakeshore Road, St Catharines 28 Dalemere Crescent, St Catharines 12 Courtland Road, St Catharines 22 Corbett Avenue, St Catharines 36 Scullers Way, St Catharines 8 Pawling Street, St Catharines 22 Johnston Street, St Catharines 13 Bluespruce Court, St Catharines 27 Prince Paul Crescent, St Catharines 32 Ernest Street, St Catharines 511 Vine Street, St Catharines 37 Ghent Street, St Catharines 540 Geneva Street, St Catharines 19 Chatham Road, St Catharines 413 Niagara Street, St Catharines 10 Champa Drive, St Catharines 473 Scott Street, St Catharines 363 Bunting Road, St Catharines 13 Gormley Crescent, St Catharines 1A SunnyLea Drive, St Catharines 3 Ennismore Court, St Catharines 1451 Fifth Street, St Catharines

644084.73 644635.60 637485.61 638997.96 639311.53 639702.77 639679.00 640042.62 6396221.34 640064.39 641655.93 642088.98 641472.57 643676.59 642974.03 642821.44 644819.32 644148.39 644394.53 644605.89 645439.79 645269.55 645079.26 646032.96 638227.79

4784853.23 4784870.30 4782754.12 4783300.55 4783366.76 4782845.76 4784137.55 4783692.17 4784235.33 4784119.82 4784010.92 4784237.02 4782888.61 4784374.65 4783323.08 4784247.01 4783147.20 4783197.67 4783701.95 4783658.68 4782888.95 4782862.16 4782743.87 4782894.36 4781917.22

Geo Tek | Geostatistical Report

March 22, 2013

18 18 19 19 19 19 20 20 20 20 21 21 21 21 22 22 23 25 25 25 26 26 26 27 27

Martindale Martindale Martindale Martindale Martindale Orchid Park Orchid Park Orchid Park Fitzgerald Orchid Park Facer Facer Facer North End Bunting Bunting Louth Martindale Martindale Martindale Haig Haig Haig Fitzgerald Fitzgerald

$350,000.00 $639,900.00 $279,000.00 $166,000.00 $219,900.00 $214,900.00 $269,900.00 $177,000.00 $169,900.00 $132,000.00 $229,900.00 $164,900.00 $159,900.00 $117,500.00 $190,000.00 $229,900.00 $629,900.00 $629,900.00 $149,900.00 $359,900.00 $119,900.00 $259,900.00 $174,900.00 $77,900.00 $226,900.00

56 Henley Drive, St Catharines 1 Brooklyn Court, St Catharines 70 Scott Street, St Catharines 104 Ventura Drive, St Catharines 118 Haig Street, St Catharines 11 Fonthill Court, St Catharines 11 Kingsway Crescent, St Catharines 11 Hill Park Lane, St Catharines 21 Sandown Street, St Catharines 222 Carlton Street, St Catharines 68 Parkview Road, St Catharines 50 Parkview Road, St Catharines 54 Cosby Avenue, St Catharines 110 Garnett Street, St Catharines 24 Huntley Crescent, St Catharines 15 Rendale Avenue, St Catharines 1665 Gregory Road, St Catharines 40 Tulip Tree Common, St Catharines 6 Barton Street, St Catharines 5 Inglis Circle, St Catharines 153 Pleasant Avenue, St Catharines 44 Chicory Crescent, St Catharines 21 Taylor Avenue, St Catharines 173 Vine Street, St Catharines 56 Maple Street, St Catharines

640455.75 639889.22 646839.12 641388.35 641521.65 641979.42 643354.16 643412.50 643525.30 642801.30 644413.60 644454.89 644376.99 644587.06 645907.72 646017.00 637008.61 640599.05 640599.53 640144.49 642095.95 641354.83 641891.45 643841.79 643011.45

4782190.58 4782250.50 4784794.59 4781674.62 4781748.97 4782457.18 4781780.11 4782160.41 4781880.17 4781675.35 4781881.45 4781775.37 4781807.82 4782301.41 4782612.61 4782802.31 4780897.32 4781227.53 4780431.14 4779930.68 4781014.32 4781289.07 4780930.48 4781353.66 4780915.30

Geo Tek | Geostatistical Report

March 22, 2013

27 27 28 28 28 28 29 29 29 29 30

Fitzgerald Fitzgerald Queenston Queenston Queenston Queenston Kernahan Queenston Queenston Kernahan Lock 3

$154,900.00 $154,900.00 $99,794.00 $204,900.00 $155,000.00 $159,900.00 $89,500.00 $224,900.00 $179,900.00 $160,000.00 $538,800.00

42 McGhie Street, St Catharines 59 Vine Street, St Catharines 30 Parkview Road, St Catharines 62 Chelsea Street, St Catharines 25 Berryman Avenue, St Catharines 17 Berryman Avenue, St Catharines 23 Emmett Road, St Catharines 97 Bunting Road, St Catharines 25 Lorne Street, St Catharines 31 Emmett Road, St Catharines 15 MacKenzie King Avnue, St Catharines 2098 Seventh Street, St Catharines 66 Elderwood Drive, St Catharines 53 West Farmington Drive, St Catharines 74 Sawmill Road, St Catharines 228 First Street, St. Catharines 29 Yates Street, St. Catharines 14 Trafalgar Street, St. Catharines 10 Norris Place, St. Catharines 55 Yates Street, St. Catharines 63 Glenridge Avenue, St. Catharines 54 Highland Avenue, St. Catharines 23 Hainer Street, St. Catharines 43 Highland Avenue, St. Catharines

642705.71 643870.98 644430.96 644804.36 644171.39 644178.28 646401.30 645575.73 644976.29 646418.58 646576.55

4781574.66 4780723.05 4781628.54 4780530.40 4780463.24 4780425.61 4780417.82 4780266.74 4780278.76 4780362.62 4780191.10

31 33 33

Louth Vansickle Vansickle

$649,000.00 $309,900.00 $435,000.00

637672.50 640673.75 640468.37

4778582.77 4779794.27 4779775.04

33 33 34 34 34 34 35 35 35 35

Vansickle Vansickle Western Hill Western Hill Western Hill Western Hill Western Hill Western Hill Western Hill Glenridge

$389,900.00 $359,900.00 $549,900.00 $649,900.00 $749,900.00 $1,350,000.00 $289,900.00 $374,800.00 $149,900.00 $650,000.00

641140.93 640108.41 642391.57 642447.13 642152.28 642104.50 643105.28 643297.83 642523.17 643290.53

4779810.22 4779939.54 4779453.21 4779513.39 4779680.03 4779564.31 4779041.31 4779404.20 4779029.04 4779546.38

Geo Tek | Geostatistical Report

March 22, 2013

36 36 36 37 37 37 37 38 39 39 41 41 42 42 42 43 43 43 43 44 44 44 45 45 45

Glenridge Glenridge Oakdale Secord Woods Oakdale Secord Woods Secord Woods Secord Woods Louth Louth Vansickle Vansickle Vansickle Western Hill Vansickle Western Hill Western Hill Western Hill Glenridge Glenridge Glenridge Glenridge Secord Woods Merritton Merritton

$139,000.00 $239,900.00 $189,900.00 $357,500.00 $132,900.00 $174,900.00 $159,900.00 $329,900.00 $1,995,000.00 $569,900.00 $349,900.00 $359,900.00 $114,000.00 $224,000.00 $165,000.00 $254,900.00 $399,999.00 $449,900.00 $369,900.00 $275,000.00 $379,000.00 $144,000.00 $219,500.00 $103,900.00 $349,900.00

6 Phelps Street, St. Catharines 2 Marren Street, St. Catharines 298 Oakdale Avenue, St. Catharines 28 Woodrow Street, St. Catharines 27 Battersea Avenue, St. Catharines 37 Greenwood Avenue, St. Catharines 52 Greenwood Avenue, St. Catharines 3 Alex Grant Place, St. Catharines 3420 Ninth Street, St. Catharines 1673 St. Paul Street, St. Catharines 15 Consiglia Drive, St. Catharines 52 Strada Blvd., St. Catharines 48 Church Hill Street, St. Catharines 28 Cumming Street, St. Catharines 35 Lloyd Street, St. Catharines 45 Rivercrest Drive, St. Catharines 111 South Drive, St. Catharines 47 Hillcrest Avenue, St. Catharines 25 Riverview Blvd., St. Catharines 71 Village Road, St. Catharines 27 Adelene Crescent, St. Catharines 168 Oakdale Avenue, St. Catharines 16 Rampart Drive, St. Catharines 20 Chestnut Street, St. Catharines 368 Merritt Street, St. Catharines

644589.87 644266.91 644029.78 645790.40 645473.22 645849.50 645943.29 646621.41 636907.27 637937.85 640886.08 641083.64 641797.25 642267.18 641915.16 643041.78 643231.63 642967.70 642888.28 643832.64 643763.57 644579.04 645808.10 645265.43 645263.62

4779059.90 4779394.41 4779967.80 4779806.49 4779559.71 4779719.24 4779720.40 4779921.26 4777505.39 4777679.81 4777726.63 4777791.16 4778145.97 4778061.14 4778243.01 4778346.23 4778819.56 4778852.48 4777800.39 4777690.53 4778099.34 4778896.22 4778756.05 4777841.19 4778087.06

Geo Tek | Geostatistical Report

March 22, 2013

49 49 49 50 50 50 51 51 52 52 52 53 53 53 54 54

Vansickle Vansickle Vansickle Powerglen Riverview Powerglen Riverview Marsdale Brockview Glenridge Glenridge Merritton Burleigh Hill Burleigh Hill Merritton Merritton

$545,000.00 $186,900.00 $369,800.00 $99,900.00 $599,000.00 $334,900.00 $349,900.00 $389,000.00 $389,000.00 $254,900.00 $234,900.00 $725,000.00 $132,900.00 $194,900.00 $175,000.00 $279,900.00

68 McCaffery Crescent, St. Catharines 218 Rykert Street, St. Catharines 93 McBride Drive, St. Catharines 198 Pelham Road, St. Catharines 280 Riverview Blvd., St. Catharines 259 Pelham Road, St. Catharines 2 Parklane Crescent, St. Catharines 12 Valerie Drive, St. Catharines 6A Pearl Ann Drive, St. Catharines 16 Glengarry Road, St. Catharines 10 Brookdale Avenue, St. Catharines 46 Ridge Point Drive, St. Catharines 68 Queen Street, St. Catharines 64 Rose Street, St. Catharines 54 Welland Avenue, St. Catharines 57 Welland Avenue, St. Catharines

640915.76 640805.07 640843.90 641973.92 642970.97 641759.54 642717.47 643591.71 644300.64 644756.71 644859.36 645174.02 642267.02 640824.33 642151.79 642205.43

4776246.53 4777428.80 4777354.35 4777220.88 4778096.75 4776895.80 4776976.48 4777061.70 4776809.85 4777608.79 4777633.17 4776881.33 4779903.42 4780720.23 4780291.92 4780310.28

B.1 Raw Data of Neighborhoods, House Prices, Addresses and Coordinates in St. Catharines

Geo Tek | Geostatistical Report