A few problems and errors were encountered during the process of my analysis which may have skewed the overall result. The first error was that the 1996 census data, being the most current, is dated. There has been tremendous growth between the last census and the present which was not taken into consideration for this project. My next problem was discussed earlier when I said no street names were available for the roads layer. Also, the resolution after converting the roads layer from vector to raster was worse than before. This made digitizing of points for existing shops more difficult and less accurate as they were approximated on screen. This created distortion as to their exact locations. 

My biggest problem was that I couldn't run the MCE when all my layers were on different projections. Something I had failed to realize until the end. I had to go back and make all the layers the same projection as my commercial layer which I decided to use as the standard. Here is the projection of my commercial layer.

Columns: 1520

Rows: 1271

Min X: 468600

Max X: 544600

Min Y: 5427650

Max Y: 5491200  

From here, for each layer in IDRISI, I used reformat and project that showed previous output numbers which were then changed to the standard commercial layer output numbers above. Then, I added fix to my newly saved layers.

(1) combufferfix.rst

(2) incomefinalfix.rst

(3) popfinalfix.rst

(4) roadsfinalfix.rst

(5) skybufferfix.rst

(6) shopbufferfix.rst

After that was done, I was able to run my MCE smoothly. It is important to note that this project was a crude examination of deciding where suitable locations would be for a new snowboard shop. There would be many more factors and constraints involved and the hard boolean approach is not always the best method. The boolean MCE analysis using logical AND (minimum) operation, in terms of risk is very conservative. Only by exactly meeting all criteria is a location considered suitable. This is a limitation and a non-realistic approach. My lack of experience working with IDRISI prevented me from attempting to do the Non-Boolean Weighted Linear Combination (WLC) approach which is an averaging technique used to place an analysis halfway between the AND (minimum) and OR (maximum) operations. It would have been a lot more accurate in terms of a final result.

<< back | next >>

main | conceptual outline | data collection | methodology

spatial analysis | problems & errors | about the author