VECTOR-TO-RASTER conversion of Housing Market Data :

Vector GIS is often used to define the urban geographic surface by points, lines and polygons, on the assumption that business-geographic information is commonly collected and analysed that way.  Each place on the earth's surface can lie on or near a point (eg:  a city, an apartment building), on or near a line (eg:  a street) and/or within a polygon (eg:  a census tract, or rental market zone).  Each of these vector entities have attributes (eg: street address, average rent, vacancy, hydro expense) associated with them, stored in a database that is linked to the geographic objects.  A rental market zone, as a GIS polygon, can itself have attributes associated with it, such as the "average rent" taken from all the apartment buildings (points, with rent as their individual attributes) within that polygon, or rental market zone.

Raster GIS defines the geographic surface as rows and columns of pixels.  There are no points, lines or polygons.  Instead, think of the earth's surface as a giant "spreadsheet", of columns and rows.  Each "cell" on that spreadsheet (or "pixel" in the raster GIS image) has an address, such as "D41" and contains a single bit of information, such as telephone expense.  Each cell is exactly the same size, and no part of the spreadsheet is without a cell, even though many may be empty.

A raster GIS represents the earth's surface the same way.  Instead of rental market zones as polygons in a vector GIS, the metropolitan area is now a "sea" of pixels, each of which stores a value for rent, each of which can be referenced in the way a cell in a spreadsheet can.

A vector entity, such as a point representing an apartment building, or a polygon representing an rental market zone, can have numerous attributes associate with it.  For instance, clicking on rental market zone 5 in Calgary's vector GIS can bring up the average one-bedroom rent of the past 6 years and the current growth rate of rental households in it.

Each raster GIS pixel, on the other hand, stores only one value.  It may be rent, it may be population growth rate.  However, that value can be manipulated, re-calculated analysed, summarised, etc.  As well, one raster image storing rent information can be overlain onto another of the same rental market zone storing population information, to satisfy "IF-THEN" types of queries (these are known as "Boolean overlays", since they use standard Boolean algebraic logic equations).  One can isolate neighbourhoods where both the rent and population are dropping most rapidly, for example.  Certainly this is also possible in a vector GIS, but with not nearly the same level of accuracy nor simplicity.

In fact, one major advantage of raster GIS is that it lends itself to complex mathematical analysis not possible in vector GIS.  For instance, trend surface polynomial expressions can be calculated (ie: multiple regression equations, where the independent variables are latitude & longitude, a common tool in economic geography) that allow one to create models of geographic behaviour, and predict unknown values based on known inputs.

The images below demonstrate how rent may be stored in a raster GIS.  They were derived first by address-geo-coding a Lotus 123 spreadsheet, containing the addresses of apartment buildings in a Canadian city, with the year they were built and their rents, onto a street-network file (SNF), in MapInfo.  Next, an SQL query was used to isolate one-bedroom rents in buildings constructed between 1960 and 1970.  The following steps describe the hybrid process of vector-to-raster conversion :

resulting in a continuous surface of average rent.  The following images illustrate this step by step evolution from vector to raster.
This is the "source" vector point data.  Note the sample of average one-bedroom rents associated with different apartment buildings.

The next image represents one way that the above vector point data may be represented by vector polygons:

The following 4 images represent different possible raster GIS images resulting from interpolating the above vector point data.  The apartment buildings (vector points) from the 1st image above, as "source" data, are overlain to give perspective.  Each of these images contain ~60,000 pixels, and each pixel has with a "rent" value associated with it:

Your comments on this page are welcome!

Back to Contents