Operational Errors

The first error I encountered with creating my shapefile was the order I had listed the feature-latitude-longitude in excel because my points did not show up on ArcMap no matter how I attempt to change the projection. The solution to this problem was to change the order to latitude-longitude-feature because now ArcMap has no problems reading my x,y coordinates first and then attributes last.

To ensure data is correctly projected; all shapefiles must have the same geographic coordinate system (GCS_Clarke_XXXX_ and datum (NAD_1983_zone_10_N) otherwise positional errors will continue to propagate. Once all shapefiles are correctly projected and ready to undergo "feature to raster" function we must manually change the pixel size (I picked 20) for all our files that are to be rasterized because if pixel size are not the same they cannot be overlay on top of one another to perform map algerba.
By selecting "clip" function from Data Management Tools --> Raster and specifying the min, max of X and min, max of Y we ensure that the same area will be clipped. Once data is imported to IDRISI check the layer properties to make sure the cell resolution are consistent (for my analysis: c1410 x r1264)

Data was loss during the process of converting from feature to raster from ArcToolBox when dealing with census data. Due to the size of some the DAUID (dissemination area units) being very small, during rasterization of these extremely small polygons they are not recongized as a single pixel in raster format because they became dissolved to their neighbouring pixels. So if analyzing very small areas, user must be aware of the loss of important but tiny polygons.

Another problem encountered was when I attempted to import from ASCII (*.txt) to IDRISI (*.rst) I could not select to have all numbers converted from real to integers because an error 15707 would pop up. At first, I thought it my text file or my shape files that had the problem so after many attempts to rasterize and convert to ASCII I realize the root of the problem lies in the "convert output file from real to integer" must NOT be checked off. If you convert everything to integers such as population density then the values greater than 30,000 will simply be not exported in IDRISI from ASCII to raster. Then the analysis in IDRISI would be incomplete. If users are not aware of their dataset this could easily slip through and cause error propagation as it is not visible and will not result in any operational errors.

Methodological Errors

While Pairwise comparison in IDRISI can generate weights by assigning rating ratios between user defined factors, the user is cannot assign ratings according to their standards because the result of user freely defining ratings are unacceptable consistency ratios. The results were as high as 0.23 during my first attempt to assign weights which is unacceptable because CI should be 0.09 or less. Essentially, for the CI to be at acceptable levels, the user can only set the first set of ratings for any one row or column and the rest are pre-defined. So in essence, I was not able to freely assign ratings because that would result in unideal consistency ratios for any type of analysis.