To determine locations in Vancouver suitable for the development of a luxury, assisted living home for seniors, a multi-criteria evaluation (MCE) was performed using a non-boolean standardization and weighted linear combination. Before this evaluation however, the newly rasterized data required some manipulation in order to display the desired variables. The following is a step by step of the methods used and can are visually illustrated by the cartographic model.

The Vancouver image was reclassified from cell values that indicated distinct enumeration areas to a boolean image with cell values that indicated whether the cell was located within Vancouver or not. Parks and Electoral District A had a 0 value as they are not covered by Vancouver enumeration areas.

In order to create images that displayed the desire census data, Attribute Value Files (AVL) files were made by using Database Workshop and exporting the required columns (attributes) from the attribute table. These AVL files were then "ASSIGNed" to the Vancouver enumeration area image to produce images with the desired data for each distinct enumeration area.

In order to perform a MCE that produces logical results, the input images must have the same range of values. For this analysis, the range of values used was 0 - 255 to allow for continuity along the suitability scale. The three images displaying population data were RECLASSified using the same parameters (see cartographic model) so that their values fell within this range. The parameters chosen are subjective and based on a visual analysis of the data and gut instinct.

For the parks, seawall, bus stops, skytrain, amenities, and homes images, it was not the existance of these phenomenon that were of significance but rather the distance from them. Using the DISTANCE module, each cell of these images took on values that indicated their distance from the closest feature. This operation resulted in a range of values for each image that was then standardized by using the FUZZY module and choosing the linear, monotonically decreasing options (except for "homes" where, as distance increased so did suitability - this used the linear, monotonically increasing option).

In addition to the boolean constraint image already produced of Vancouver, the "dwelling value" image was also reclassified as a boolean, constraint image that indicated whether cells fell within enumeration areas where the average dwelling price was $500,000 and greater or not.

The WEIGHT module was used to assign relative weights for each of the above mentioned criteria so that more suitable criteria can compensate for less suitable criteria in the final analysis. The weights chosen result from the ranks given each criteria (in conceptual outline) and can be viewed at the bottom of the cartographic model.

An MCE was performed using the weighted criteria and two constraint images. The resultant image shows the relative suitability, on a scale of 0 - 64, of each location within Vancouver and within enumeration areas where the average dwelling price was $500,000 and greater. The cells that fell within the top third of these values were isolated by RECLASSing the image.This image showed only areas that scored 40 or better in the MCE.

I arbitrarily decided that only suitable areas of 1 hectare or greater would be useful for a new development. To isolate these areas, the GROUP module was used to assign unique values to groups of pixels that were contiguous. Because their were "islands" of unsuitability within suitable areas, these groups of cells also got unique values thus producing several values that indicated unsuitability. In order to regain just one value for unsuitable areas, an AVL was produced using Edit and then ASSIGNed to the grouped image "suitability group". The resulant image was named "suitability group 2".

In order to determine the area in hectareas of contiguous suitable locations, the AREA module was used with the "suitability group 2" image. This produced an image where cell values now indicated the area of their group with the largest value being that of the unsuitable area. A boolean image was also created from "suitability group 2" to show areas that were suitable or not. By OVERLAYing these two images using the "multiply"option, the unsuitable areas were reassigned a value of 0. This is important because the next step was to RECLASSify the image to show only suitable groups that have areas over 1 hectare. Had the OVERLAY operation not been performed, unsuitable areas would have been included.

Finally, the resultant map shows areas within Vancouver that are suitable for development of a new assisted living seniors home based on the chosen criteria. In order to put these areas in context, a main street shapefile and shoreline file were imported and converted to a IDRISI vector format. These vector files were then added on top of the raster image to produce the final image.