Massachusetts Institute of Technology
Department of Urban Studies and Planning

11.520: A Workshop on Geographic Information Systems

11.188: Urban Planning and Social Science Laboratory

Homework 1: Mapping of Community Characteristics

Due (online) on Wednesday, October 13, 2010, before the start of class - 2 PM.

Download the PDF version of Homework 1.

NOTE: Homework assignments will take longer and be much more important to your grade than any one lab exercise, so devote your energy accordingly.

INTRODUCTION

In this exercise, you will explore the spatial patterns of the housing and socio-economic characteristics of communities in and around Boston . To assist in this task, we provide: (1) demographic data at the census tract level from the 1990 US Census, and (2) boundary files for cities and towns, major roads, shopping center locations, and census tracts.

We ask you to use these data to prepare a short report with textual discussion plus three maps for Part I and two maps (with a table) for Part II.

Before starting the hands-on work, read through the entire assignment to get a sense of the datasets, analytic approach, and processing steps. Then, make sure you can access the datasets in the 11.520 Locker. [As explained in the lab exercises, you can find the locker by navigating through the sub-directories on network drive Z:, or you can attach the class locker to a new drive letter - we suggest Drive M: (for maps). The earlier lab exercises explained how to mount the Andrew File System locker //afs/athena.mit.edu/course/11/11.520/ on network drive M:]

DATA

  • Census Tract Boundaries

The census tract boundaries are saved in a 'shapefile' that contains only the boundary geometry and a few geographic identifiers (like county, track number, etc.). This shapefile must be joined with a dbf-formatted table in order to relate the census data to specific census tracts. The census tract shapefile is called msa5_tr90.shp. It is located in the class data locker (M:\data) and contains all the 1990 census tracts in the five Eastern Mass counties in and around Boston.

  • Socioeconomic Data for the Census Tracts

The socioeconomic data for these tracts have been pulled from the 1990 Census SF3A datasets and are stored in the same M:\data directory in a dbf-formatted table called msa5_tr90_data.dbf. This file must be 'linked' or 'joined' to the attribiute table for the tract boundaries by a common field called "STCNTYTR" before you will be able to generate thematic maps using the census data. (STCNTYTR is the abbreviation for STate-CouNTY-TRact.) Use the ArcMap help files to see how to 'join' the data table to the attribute table if you want to get started with the homework before we show you how to do this in class.

The msa5_tr90_data.dbf table includes 60+ variables from the much longer list of all variables in the decennial Census. Take a look at the dictionary for the specific census data fields in  msa5_tr90_data.dbf. (Note: this list is a subset of the full Census Bureau's listing and technical documentation for the hundreds of population and housing variables from the 1990 census. This technical document is archived in the class locker as M:\data\census90\census90stf3td.pdf. More details about the 1990 US Census are available in Tom Grayson's and Annie Thompson's notes on "Making Sense of the Census". Be aware, however, that some of the online references in these notes are no longer available and you will need ONLY the shorter list of 60+ variables mentioned above in order to do the homework. During the next few weeks we will have additional exercises using US census data, but we will use the 2000 US census for those exercises and will provide the relevant technical documentation. Also, note that small-area comparison of census data across decades can be tricky due to changes in census tract and block group boundaries. MIT's Rotch Library has CDs from a third-party firm, GeoLytics, which has reconciled past census data to year 2000 boundaries. Should you wish to analyze trends in census data for your individual project later in the semester, you will likely want to use the GeoLytics CD.)

  • Shopping Centers and Major Roads

Besides the census data, which will be used primarily in Problem #1, you will need a map of major roads and shopping centers for Problem #2. The shopping center coverage (for the Boston metro area) is called shopcntrs and is also stored in M:\data. The major roads layer is called majmhda1 and can be also found in the directory M:\data. All of these coverages use the following coordinate system: Massachusetts State Plane, Mainland Zone, NAD 1983, meters. **Be sure to set the map units and distance units in the Data Frame Properties window so you can measure distances in your Data View window and be sure that the distances you compute in Problem #2 are reasonable.**

Data Sources: The roads coverage comes from the Mass Highway Department via MassGIS and the shopping center coverage is proprietary data provided by SSR Research (circa 1995) for internal MIT educational use.

SUMMARY

A map should always have a purpose. A good map should deliver the information that you want readers to understand. Therefore the map should be very intuitive without requiring reading the discussion of the map in your paper or report. Try to give the map to your friends who have no training in GIS to see if they can recognize the message you were trying to deliver and ask them whether they find the evidence to be compelling.

  • Your goal is to gain some understanding of housing and socio-economic patterns in metro Boston.
  • Try to have fun and explore ArcMap while doing this homework. Use the on-line help to experiment with ArcMap's capabilities.  Here's another mapmaking hint: you can specify different colors for your foreground and background features, you may want to turn off the outlines of polygons in a layer, and you may want to choose different widths and sizes for lines and symbols. Exploit this functionality when overlaying different data layers.
  • To make your maps more easily interpreted, you can overlay the political boundaries of Massachusetts towns. Also consider using the water bodies in Eastern Massachusetts. The Mass boundaries are located in M:\data\matown00.shp. A shapefile of Eastern Massachusetts water bodies is available in M:\data\msa_water.shp.

Problem 1: Exploratory Mapping
Metropolitan Area Census Data (60 points)

1. [20 points] Create a thematic (or chloropleth) map showing the population density of the MSA.

  • You should calculate density as persons (P0010001) per acre (landacre). Normalizing by the 'landacre' variable in the census data is more reliable than using the 'area' variable since the tracts extend into Boston harbor, the Charles River, etc. whereas the 'landacre' variable is the census estimate of land acreage within each Tract.
  • Be sure to include only those polygons (census tracts) where landacre > 0 and the relevant census data are not missing. (For example, census data are missing from a cluster of tracts north of Boston.) Use the Select > Select by Attribute menu option to select the tracts you want to include.
  • Classify the data into approximately five categories.  The classification method you use is up to you, and can include customized category breaks.  Experiment with the available methods and pick the one you feel best clarifies your exploratory interest (for example, do you want to differentiate within high density areas, or characterize the full range of densities?)   In a sentence or two, explain your focus, your choice of classification, and why your cartographic technique makes sense.

2. [20 points] Map the homeownership ratio--the ratio of owner occupied housing units to the total occupied housing units.

  • Remember that the 'tenure' variables count the number of owner-occupied and renter-occupied housing units.  Be careful about what the numerator is and what the denominator is.
  • Just as for the previous map, you will also need to exclude tracts that lack adequate data. Again, include a brief few sentences explaining your choice of classification scheme and any pattern that you want to show.

3. [20 points] Map another Census attribute of your choosing with interesting spatial patterns using the same process as described in number two.

  • For all three maps, be sure to include a title, source, logo, north arrow, scalebar, and legend (with indication of classification choice). The quality of your presentation does count!
  • In separate text explain the "story" that your map tells. What spatial patterns, if any, are suggested? (Don't try to over explain the map!) Explain the reasoning behind your classification choices, how you handled missing values, and any other relevant judgments and assumptions. Please limit this discussion to a total of one page.  We do not want lengthy explanations. There is more than one reasonable choice for your classification and normalization choices. We simply want to be sure that you've thought about the issues and made reasonable choices.

Problem 2: Introductory Spatial Analysis 
Relationships between Roads, Shopping Centers, and Residences

For this problem, you are asked to investigate the relationships among the locations of shopping centers, major roads, and residential clusters. After doing some exploratory mapping as you did in the first problem, you are asked to dig a bit further into the data, develop a few specific measures that carefully exclude incomplete or inapplicable data, and then develop maps that successfully visualize the results and the reasoning behind your analysis.

The shopping center data are stored in a shapefile at M:\data\shopcntrs.shp. (These data are proprietary and not to be used or redistributed for non-MIT purposes.) Included in these data are characteristics such as square footage of retail space (totalsf as a text field and squarefeet as a numeric integer) and type of center (propertysu).

Explore these two variables to try to determine if a relationship exists between them. To do this you may want to calculate the average size of each type of shopping center. Note that not all observations include a value in the totalsf or squarefeet fields. Note that these shopping 'centers' do not include places like Central Square (Cambridge) where commercial/retail activity is present among individually owned parcels and buildings along a city street. This dataset focuses on shopping center developments where a large tract of land or strip mall under common ownership is divided up into clusters of businesses.

1. [20 Points] Create one map showing the relationship(s) between shopping center location and the location of major roads and population centers.

  • Be sure to use different symbols and/or sizes on your map for the different types of shopping centers. Likewise differentiate major and not-so-major roads -- use the class field in the majmhda1 attribute table.
  • Use high population density as an indicator of where population centers are located and shade the tracts based on population density (as you computed it in problem 1.
  • Be sure to turn the tract outlines off so they don't clutter the map. In fact, it will take some effort to develop maps with good symbolization and cartographic choices so that they are both readable and informative.

2. [20 Points] Buffer the major roads and create a second map that examines whether certain types of shopping centers tend to be inside the buffer.

  • Select the Interstate Highways and Routes 2 and 3 (not 3A) from the major roads layer.
  • Use the buffer tools to create an 800 meter buffer around these selected roads and then calculate the share of shopping centers by type (among those within the 5-county region) that lie within the buffer.
  • Map your results and include in your "layout" a map of the entire area, as well as a more detailed map zoomed into an area of the region with interesting spatial patterns.
  • In addition to your two map views, create and display a table showing, for each type of shopping center (propertysu), the number of such shopping centers within the 5-county msa5_tr90 region, and the number and percentage of each type of shopping center that fall within the buffer areas.

Explain in a couple of paragraphs, separate from the maps, (a) the steps you took to select those roads and shopping centers that you included when computing your statistics, and (b) your interpretation of any general pattern that you observe regarding the location of shopping centers, major roads, and population centers. In particular, be sure that your discussion covers:

  • What conceptual relationships are these maps intended to portray?
  • Which classification schemes did you choose to use, and why?
  • Which shopping centers and tracts have you excluded from the analysis, and why?
  • Do there appear to be any interesting spatial relationships shown on the map?

Homework Requirement

Don't just turn in the maps! You should turn in a short report that integrates the maps and tables together with the explicit answers to both questions. Use the maps and tables in the paper to illustrate and amplify your verbal reasoning rather than simply to produce maps without a stated context and purpose.

Please submit your homework (in WORD, RTF and/or PDF format) using the Stellar homework turn-in capability at http://stellar.mit.edu/S/course/11/fa10/11.520 or http://stellar.mit.edu/S/course/11/fa10/11.188

by 2 PM Wednesday, October 13, 2010.

Created and Modified:1993-2010 by Raj Singh, Thomas H. Grayson, Annie Kinsella Thompson, Joseph Ferreira, Myounggu Kang, Jschung Chung , Jinhua Zhao, Mike Flaxman, Yi Zhu, Lulu Xue, and Shan Jiang

Last modified: 4 October, 2010, [jf],[shanjang]

Back to the CRON Home Page.
Back to the 11.520/11.188 Home Page.