add an exponential normalization option to linkage Priority tools

[johngallo]

currently, there is the lp_settings.py option to normalize using score range or max value.

Relative closeness and size normalizations could benefit greatly from an inverse exponential, or even a "histogram equalize" option, so that very short linkages are not off the charts compared to regular length and long linkages.

Rather than the following values be in linear:

4, 9, 16, 25, 36...

2,3,4,5,6...

Cherry on top is to define the exponent as the second paramater. (in this case it is 0.5)

Looks fairly straightforward.

In the below script, add

NM_SQR = "Square Root" # Square Root Normalization

and add an elif (else if) and the function?

    if normalization_method == NM_SCORE:
        if invert:
            return (max_val - in_raster) / (max_val - min_val)
        return (in_raster - min_val) / (max_val - min_val)
     elif normalization_method == NM_SQR:
        outSQRT = SquareRoot(in_raster)

{Code needs to here to get the new_max_val and new_min_val raster after the square root, and that are needed for the next step (normalizing linearly to 0-1)}

        if invert:
            return (new_max_val - outSQRT) / (new_max_val - new_min_val)
        return (outSQRT - new_min_val) / (new_max_val - new_min_val)
        # double check the math logic of the above and that if inverting you sholdn't square it first...

    else:  # Max score normalization
        if invert:
            return (max_val + min_val - in_raster) / max_val

XXXXXXXXXXXXXXXXXXXX

_SCRIPT_NAME = "lp_main.py"

NM_SCORE = "SCORE_RANGE" # Score range normalization
NM_MAX = "MAX_VALUE" # Maximum value normalization

CoordPoint = namedtuple('Point', 'x y')

class AppError(Exception):
"""Custom error class."""

pass

def normalize_raster(in_raster, normalization_method=NM_MAX, invert=False):
"""Normalize values in in_raster.

Normalize values in in_raster using score range or max score method,
with optional inversion.
"""
lm_util.build_stats(in_raster)
result = arcpy.GetRasterProperties_management(in_raster, "MINIMUM")
min_val = float(result.getOutput(0))
result = arcpy.GetRasterProperties_management(in_raster, "MAXIMUM")
max_val = float(result.getOutput(0))
if max_val > 0:
    if normalization_method == NM_SCORE:
        if invert:
            return (max_val - in_raster) / (max_val - min_val)
        return (in_raster - min_val) / (max_val - min_val)
    else:  # Max score normalization
        if invert:
            return (max_val + min_val - in_raster) / max_val
        return in_raster / max_val
else:
    return in_raster * 0