Home | UBC Blogs



GEOB 270 2017 Term 2Lab 4: Housing Affordability Working with Census Data: tables, joins, classification, normalization, visualizationIntroduction and Scenario Often geographers need to work with census data that is gathered, analyzed and displayed by national government organizations (example Statistics Canada). Census data is utilized in many geographic analysis research projects: urban planning, immigration, social geography, environmental justice, crime prevention, transportation planning, policy analysis, etc., and all of these topics can be explored using GIS. Every 5 years, Canadians are surveyed. Previous to the 2011 survey, 80% of the population completed the ‘short form’ and 20% of the population completed the ‘long form’, where more detailed information is captured. In particular, the ‘long form’ included socio-economic data including income, dwelling type and value (the topic of this lab), transportation data such (modes of transportation for commuting) and immigration data (as seen in the UBC Metropolis Atlas geog.ubc.ca/metropolis/, which you have seen in lectures). The former Canadian government changed census policy in 2010; the long form was not mandatory but voluntary. This has impacted the quality of the data, and makes it challenging to compare to previous years. The next survey was conducted in 2016 and the data is just being released, but so far only population changes by metropolitan areas were released last week. We will have to use 2011 data for this lab. For this lab, you will learn how to access and download Canadian census data, and review the spatial resolution for census track data (CTs). There are other types of census data, such as dissemination data (DAs) that we will not be using in this lab. You will then explore various methods for classifying quantitative data, using cost of housing as the variable for metro Vancouver CTs. Finally, you are given a scenario that you are currently employed by the Globe and Mail, Canada’s national newspaper, as a data journalist and need to compile some maps to accompany an article on cost of housing and housing affordability, which is one of the ‘livable’ city indicators. There are many indices used to measure ‘livability’ and Vancouver is thought to be one of the most ‘livable’ cities in the world. However, when looking at one metric of livability, housing affordability, Vancouver is one of the two or three most unaffordable cities in the world. You will calculate housing affordability (a ratio of household income to cost of owner dwelling) for Metro Vancouver and a smaller city near London – London Ontario. As part of the lab, you will need to read pages 7-9 of this document for a description of this ratio, and the resultant affordability classifications (). The maps you create for Vancouver and London on cost of housing and housing affordability will be used to accompany an article in the Globe and Mail.Due: 2 week lab, due at the beginning of your lab time in 2 weeks, (plus reading break, so 3 elapsed weeks Learning Objectives and TasksPart 1 - Developing a working knowledge of Canadian Census DataDownloading Spatial and Tabular Census Data (there is no ‘getdata’ for this lab, you will be accessing and downloading data from an internet site)Join tabular data to spatial layersVisualizing housing data Terms of Canadian Census Data collectionPart 2 - Understanding quantitative data classification Creating a map to illustrate the difference between 4 methods of classification: natural breaks, equal interval, standard deviation and manual breaksPart 3 - Working with ratios to compare datasetsNormalizing data to determine housing affordabilityPart 4 - Creating maps of GIS analyses results Due: 2 week lab, due at the beginning of your lab time in 2 weeks. Any labs submitted after the start of your lab time, will be marked late. Answers to lab questions will be accepted in a pdf for map images and pdf for text answers. Submit answers to questions via the course web site: up your workspaceGenerally speaking, when starting any GIS project, you: set up a workspace (c:\temp\lab)copy any data into the workspace create a geodatabase (if using ArcGIS as we are in this course) Import the data into the gdbContinue with analysisSet up a geodatabase called Lab4housingIn the ArcCatalog Catalog tree, right-click on C:\temp\lab4 and select New > File geodatabase and name it lab4housingLaunch ArcMap. In the Getting Started window, set your default geodatabase to C:\temp\lab4\lab4housing.gdb. Select the Blank Map template from My Templates and click on OK to begin. Setting your default geodatabase will help you to avoid losing your data. It will ensure that the output of any analytical operation is stored in lab4housing.gdb as a feature class, unless another location is chosen.Part 1 - Developing a working knowledge of Canadian Census DataDownloading Spatial and Tabular Census Data You will download the census data from the University of London census data site. You will find, download and extract:Spatial data – CT census boundary maps.Tabular data - for shelter dwelling cost and family household income that relates to the CTsAccessing Census map files for Canada Census spatial maps or files for census metropolitan areas (CMAs), including CTs, can be downloaded from the UBC library. Go to the Abacus Dataverse Network ( ) and search for ‘Census Canada 2011 boundary files’. The page will probably ask you to login.This finds the entry to the Census Canada cartographic boundary files. Click on link and then on “Data & Analysis” tab to download individual files. The link has access to the different census boundary units.For this lab, you need to download:gct_000b11a_e.zip which are all the CT boundaries for CMAs in Canadaghy_00h11a_e.zip which are the Great lakes, St. Lawrence river and oceansgpr_000b11a_e.zip which are all the Canadian and provincial boundariescopy the .zip files into your workspace c:\temp\lab4extract the data in the zip file to the lab4 workspace (NOT THE DEFAULT which is a level below to gda_000b11a_e)Add layers to the Lab4housing.gdbRight-click on lab4housing.gdb > Import > Feature Class (multiple)In the dialog window, add all of the shapefiles and click OK. To avoid confusion, in ArcCatalog, delete the original shapefiles before continuing (all the files outside of the geodatabase). You should only be left with the geodatabaseopen ArcMap and add the CT layer and the Canada layers. Rename the layer name for gct_000b11a_e to CanadaCT (gently left click on layer name until you are allowed to edit name)Zoom to Metro Vancouver and view the CT layer (note: this will take some time to draw because this is entire Canada dataset).Extracting out Metro Vancouver and LondonBecause of the volume of data, it will be easier for you to extract into separate layers metro Vancouver and London. Open attribute table for CanadaCTYou will click on select by attribute icon1412875981075 and query for CTs with a CMANAME of Vancouver (CMAs are Census Metropolitan Areas – a clustering of municipalities such as Vancouver, Richmond, Delta…) that form large urban area) double click on CMANAMEsingle click on =click on get unique valuesdouble click on Vancouverthe query build window should look like this:applyYou should have 457 CTs selected and see them highlighted in blue around metro Vancouver. Now you need to save this selection into a layer called VanCTright click on Canada CTclick on selection – create layer from selected featuresin your listing of layers, you will see Canada CT Selection – rename this to VanCT.Turn off all layers except VanCT, zoom. Repeat this process with Canada CT (remember to clear the selection first) to extract London. query for CTs with a CMANAME of London double click on CMANAMEsingle click on =click on get unique valuesdouble click on London your query builder should look like this:Follow the rest of the steps you did for Vancouver and Rename selected layer LondonCTYou should now have these spatial layers: CanadaProv; CanadaWater;; CanadaCT; VanCT; LondonCT. Change projection (you did this in lab 2)The maps of CT boundaries are in the default coordinate system from Statistic Canada (NAD83). Since you have isolated Vancouver and London, you need to change the Canada-wide reference system to local projections. Vancouver is in UTM zone 10, London in UTM zone 17, so they need different projections.Changing projection for VancouverRight click on dataframe “Layers”, properties; change the projection to NAD 1983 UTM Zone 10Changing the projection for London Insert dataframemove LondonCT, CanadaProv and CanadaWater to dataframerename dataframe LondonRight click on London, properties; change the projection to NAD 1983 UTM Zone 17. To go between dataframes, right click on dataframe name and click activateDownloading census tabular data from University of Toronto CHASS siteFor our scenario, we need to download census data for Vancouver and London CTs for:median household income and median cost of dwelling by owner. So we are looking for: 1) household income, or family combined income of people living together in a house they own, and 2) not the average income of households but median income median (not average) cost of dwelling.QUESTION 1 (2)Why for this analysis is it better to use median cost and not average? University of Toronto has organized census data for download.Go to CHASS (*if you are working at home, you need to be connected through UBC VPN to access site). Find census profile tables by census year: 2011 NHS8890001447800Then pick profile of census tracks (cumulative) On the next page you get taken to:Step 1: Specify Census Geography for RetrievalFor geography uncheck A and check VThen check Vancouver and ensure your geography selection box looks like this:9906001384300106680014478003175001231900 Step 2 Specify NHS profile Variable for Retrieval Now you need to pick your Profile Variables for Vancouver, cost of housing and income:HousingFirst click on Hous along the topShelter cost on the vertical bars From the list of variables, check median value of dwelling 1600200469900IncomeGo back and pick inc from top menuFamilies inc in 2010 from the vertical barClick median family income check the Make sure that your selection variables look like this:276860025400Step 3 Specify output details and submit query Leave everything as it in optionalUnder download, click dBase (DBF) fileSubmit queryData Request SummaryClick link to download data file Note COL 0 is CT id, COL1 shelter costs and COL2 IncomeCopy .dbf into your workspace and rename VansheltincCTRepeat this process for London. Name file LondonsheltincCTAfter you have downloaded the two data tables, use the ArcCatalog tab to import them into the lab4housing.gdb. Add database files to ArcMap to the relevant Vancouver and London dataframes. Change column names Open attribute table for VanCTsheltincRight click on COL1 – properties –ALIAS and change COL1 to shelterRight click on COL2 – properties –ALIAS and change COL2 to incRepeat this for LondonsheltincCT b) Join the tabular data of census information to spatial layersYou now have to join the Vancouver and London map files to the database files of shelter cost and income. COL0 in your database files are the unique identifier for the CTs that match to the CTUID in the map files. Join the data to the maps with these identifiers:Right click on VanCT – joins and relates – joinIn the join window:1. Field (or item/variable) to be used for join is CTUID2. Table is VansheltincCT3. Field in table is COL0Keep only matching records Open the attribute table of VanCT – view fields. You should see shelter and inc at the endRepeat the join for LondonCT with LondonsheltincCT- using CTUID as the join itemJoins like this are only temporary, so, in order to save this work of the joins, we recommend that you save the Vancouver and London layers that are joined with the data, to a new layer in order not to lose the selection and the joinRight click on layerData Export datause defaults except for name – change name to VanCTjoin and LondonCTjoinc) Displaying housing data For VanCTjoin, symbolize housing cost by clicking:PropertiesSymbologyQuantitiesField is shelter (or housing cost )Use defaults d) Understanding Census DataTo answer the questions below, refer to the Statistics Canada website for information on the 2011 census, including definitions and descriptions of:Census spatial units like Census Tracts (under Geography) and census tabular data (variables/profiles) data (variables) 2 (2)Statistics Canada has rules for data (area) suppression (i.e. not publishing the results of census information in certain census areas). What are these rules?QUESTION 3 (2)What are the rules that Statistics Canada tries to follow in the delineation of CT boundaries? (point-form answers are acceptable) QUESTION 4 (2)For the VanCTjoin, how many have a “0” recorded for Median value of dwelling? These CTs are First Nations lands (called Indian Reservations in Canada). Why do they have values of 0? Part 2: Understanding quantitative data classification Creating a map to illustrate the difference between 4 methods of classification: natural breaks, equal interval, standard deviation and manual breaksFor our task of making maps of housing affordability, so far we have only prepared data. In any GIS and data visualization project, data preparation (searching, accessing, downloading, geo-referencing, verifying, editing, joining, etc.) can be 80% of the work. Now for the fun stuff…For your article on housing affordability, you will be making maps of cost of housing in Vancouver and London. You need to experiment with classifying the data before you select a classification method for your final maps which will be published. 1. Methods for classifying dwelling cost for Vancouver CTsThere are many different methods to classify quantitative data. The default method in ArcGIS is natural breaks—which looks for natural groupings in the data to come up with a classification based on the number of classes that you specify. The default number of classes is 5. You will often see results of GIS analysis that simply use these defaults. Unfortunately, natural breaks is not often the most accurate method of classifying data and is all too often used as the default. For this section of the lab, you will classify housing cost with 3 automated methods: natural breaks, Equal Interval and Standard Deviation, and then you will use Manual Breaks to add your own class breaks instead of computer generated class breaks. You will compare all of your results. We will use median dwelling cost (shelter, COL1) as our variable.a) Map 1: Natural Breaks (Default Classification, 5 classes)Insert a new DataFrameRename the dataframe housecost_NB (for natural breaks)Copy the VanCTjoin layer to the new data frameIf the layer is not projected properly, you need to go into DataFrame properties and project the coordinates to UTM zone 10Go to Layer Properties - Symbolize and click on Quantities and graduated colours. Ensure that the classification method is Natural Breaks and that the number of classes is 5Symbolize with the yellow to red colour ramp. Click on Classify and review the histogram – there are a number of values of 0 due to data suppression. These 0 values are included in the classification method as cost of housing = 0 which we know is an error. They really represent no data. We need to exclude these records from the data before we classify them. This is a common problem in mapping and data classification. Ensure when you are making and evaluating maps in the future that you check to see is the null values have been excludedExclusion:In the classification window with the histogram, click the “Exclusion…” bar. To select the 0s to exclude: Double click on Vansheltinc.COL1, single click on =, space, type in 0"Vansheltinc.COL1" = 0Click OK, OK **When working with Exclusion in symbology, there are sometime problems making this work correctly. This could be related to the join using “keep only matching records” which seems to confuse the renderer. An alternative would be to instead use the ‘Definition Query’ to select all records with COL1 higher the zero. ** b) Map 2: Equal Interval ClassificationInsert a new DataFrameRename the dataframe ‘housecost_EI’ (By slowly left clicking twice on the layer title)Copy the VanCTjoin layer to the new data frameIf the layer is not projected properly, you need to go into DataFrame properties and project the coordinates to UTM zone 10Go to Layer Properties - Symbolize and click on Classify and ensure that the classification method is Equal Interval and that the number of classes is 5. Check that the exclusion of 0s is still active (it should be).Symbolize with the yellow to red colour ramp. Look at the histogram. It should be clear why the classification method is called Equal Interval. c) Map 3: Manual BreaksInsert a new DataFrameRename the dataframe ‘housecost_MB’ (By slowly left clicking twice on the layer title)Copy the VanCTjoin layer to new data frameIf the layer is not projected properly, you need to go into DataFrame properties and project the coordinates to UTM zone 10Ensure that the classification method is Manual and that the number of classes is 5Under Break Values, insert the following breaks: 250000 50000010000002000000Your last break should automatically go to the highest value in your datasetUse the yellow to red colour ramp.Look at the histogramd) Map 4: Standard Deviation ClassificationInsert a new DataFrameRename the dataframe ‘housecost_SD’ (By slowly left clicking twice on the layer title)Copy the VanCTjoin layer to the new data frameIf the layer is not projected properly, you need to go into DataFrame properties and project the coordinates to UTM zone 10Ensure that the classification method is Standard Deviation. It will then automatically change the number of classes to 4 or 6. Think about why it automatically does this. Use the default symbology for Standard DeviationLook at the histogram of class breaksCreate a layout view of all 4 classification methods for Vancouver CTsCreate 4 maps in layout viewClick on view – layout Use the portrait view and display your 4 classification maps at a scale of 1:800,000Add a text box with the classification method used for each mapAdd scale bar, north arrowSave the map as lab4dataclass.mxdQUESTION 5 (5)Since you are a journalist, putting together maps of housing cost in Vancouver, which classification method would you choose for your audience and why? What if you are a real estate agent preparing a presentation for prospective home buyers near UBC? Are there ethical implications for your choice of classification method? This data is from 2011 – it is now 2017 – should you even be using this data? Discuss. QUESTION 6 (5)Save your .mxd, and then Export the map as a PDF called “dataclass.pdf”Create a housing cost map of London CTsInsert a dataframe, name it London Copy LondonCTjoin to that dataframeChange the projection to UTM Zone 18Create a housing cost map with the same Manual Breaks as the VanCTjoin. We need to use the same breaks, with the same colour scheme, to compare the housing costs between the 2 cities. (Because London has much lower housing costs, the last 2 class breaks for London may have no values. You may have to type twice the value of 2000000. )Create publishable maps comparing Vancouver to London CTs using Manual BreaksConvert your view to layoutUse the VanCTjoin Manual Breaks and LondonCTjoin Manual Breaks to display cost of housing. Edit the legend values (in symbology - edit labels) to round them to make them more readableInclude in your layout:Title and sub-titles LegendScaleNorth arrowSave your work to a new name: housecostVO.mxdQUESTION 7 (6)Export your map as a PDF called “housecostVO.pdf”QUESTION 8 (1)Natural Breaks is the most accurate and error free method of putting CTs into classes looking at the distribution of data – why then are we using Manual Breaks to compare two cities such as London and Vancouver? Part 3: Working with ratios to compare datasetsNormalizing data to determine housing affordabilitySo far we have looked at the absolute cost of detached houses in Vancouver and London. We now want to NORMALIZE the housing cost data by household income. People in larger cities tend to have higher incomes than people in smaller cities, so if we normalize the cost of housing by income for any city, we have an affordability index that makes it more accurate to compare cities of different sizes, geographic locations, and income brackets regarding ‘affordability’. You will need to read pages 7-9 of this document for a description of this ratio, and resultant affordability classifications (). You will create maps of affordability by normalizing house cost by income, and then classify the ratios as per the suggested affordability values in the demographia article.Open the housecostVO.mxd if it is not already on your screen. Save as “affordabilityVO.mxd”Create an affordability map for Vancouver CTsFor VanCTjoin > properties > symbology > quantities > graduated colorsValue shelterNormalization incClassify > Manual5 classesBreak Values.01 (this break is to eliminate the ‘0’ values as a result of no data values in the CT for median cost housing) 345Last value will be highest value in ratio OKChange the labels to text relative to the affordability classifications in the article Change the class colours to a colour scheme that shows affordable, and then various ranges of unaffordable Create an affordability map for London CTsFollow the same steps as you did for the Vancouver CTsCreate publishable maps comparing affordability Vancouver to London using Manual Breaks of the affordability indexConvert your view to layoutCreate a layout of London and Metro Vancouver affordability Include in your layout:Title and sub-titles LegendScaleNorth arrowSave your work to affordabilityVO.mxdQUESTION 9 (6)Export your layout as a PDF called “affordability.pdf”Submit Answers/Images to Lab QuestionsAnswers to lab questions will be accepted in a pdf for map images and pdf for text answers. Submit answers to questions via the course web site: PORTFOLIO POST (lab4)Quantitative Data ClassificationIncorporate Questions 6 (map) and 5 (explanation) into a portfolio post which reflects you knowledge of how different methods of data classification influence the interpretation of data on maps. Housing affordability Create a post for your portfolio that includes your map comparing housing affordability in Vancouver and London (Question 9) and text that responds to the following questions:What is affordability measuring, and why is it a better indicator of housing affordability than housing cost alone?What are the housing affordability rating categories? Who determined them and are they to be ‘trusted’? (You have seen in the previous map how different classification breaks produce very different visual impressions). Is affordability a good indicator of a city’s ‘livability’? ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download