Available with Spatial Analyst license.
In the Comparison pane, you selected the predefined statistics that you wanted run. Use the Explore Statistics pane to view the results from those statistics.
To see the results, in the Explore Statistics pane, click the statistics you want to see. The parameters for the statistic, a supporting plot, and interpretive text appear in the Comparison Statistics Pane, while an associated output map will be displayed in the map.
The parameters for the selected statistic can be altered in the Comparison Statistics Pane. The output from the statistic will be updated.
Predefined statistics for comparison
To help you analyze the output, in the following sections, what to look for in each statistic in each of the five functional statistical groups is identified.
Explore Input Parameters
There is one statistic in the Explore Input Parameters functional group that can be used to compare the input models.
Parameters in Models
See Parameters in Models for a description of the statistic and when to use it.
What to look for:
- Examine the spatial distribution of the final suitability and locate maps to gain a general understanding of the output.
- See if the models have the same input criteria and weights.
- Review the transformations for each criterion through the side-by-side plots.
- Compare the parameters for the resulting regions between the models to see if they differ.
Compare Similarities and Differences
There are four statistics in the Compare Similarities and Differences functional group that can be used to compare the input models.
Difference in Suitability Values
See Difference in Suitability Values for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for in the Percent Change option:
- Determine the range of the output values. The wider the range, the greater the disparity among the models. Examine the resulting plot to see if there are only a few locations with large differences.
- In the map, notice where the values are closest to either side of zero because these locations are where the models are in agreement.
- See where the greater positive values are since they indicate where Model 1 has larger suitability values relative to Model 2.
- See where the lower negative values are since they indicate Model 2 has larger suitability values relative to Model 1.
- See where the greater positive and lower negative values are since they show where the models differ the most.
- Notice any patterns where the models differ. These patterns might signify a bias to one of the input criterion or a combination or where one model is missing a criterion that is present in the other model.
What to look for in the Percent Difference option:
- Examine where the lower values are located since that is where the models are in agreement.
- Analyze if the lower numbers (where the models agree) and higher numbers (where the models differ) are clustering or if they spread out across the study area.
- If you see a pattern in the higher values, can you associate the pattern with a particular criterion?
Similar Between Models
See Similar Between Models for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for:
- Notice how many locations meet the similarity threshold indicating where the models are similar.
- Determine if the cells are grouped or fragmented across the study site.
- If the final regions were created, display them over the resulting map from the statistic. Examine if the regions correspond to the locations output from the statistic. If they do, the models agree where it matters.
Similarity Versus Differences
See Similarity Versus Differences for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for:
- Observe if the locations that are similar between the models are clustered. Are the locations that are different clustered?
- View if the locations that are similar are close to the locations that are different.
- Check if there are any patterns where the models show similarities and differences that can be linked to patterns in the transformed criteria.
Clustering of Differences (Hot Spot Analysis)
See Clustering of Differences (Hot Spot Analysis) for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for:
- Note whether the hot spots (where the models differ) and cold spots (where the models are similar) are clustering or fragmented across the study area.
- Inspect if the hot and cold spots are near each other or are they far apart.
- If you ran the Similarity Versus Difference statistic, compare the two outputs. Determine if including other cells around each location in the calculations (as done in the Hot Spot statistic) alters any patterns.
Analyze Suitability Values
There are four statistics in the Analyze Suitability Values functional group that can be used to compare the input models.
Models Similar in High Suitability
See Models Similar in High Suitability for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for:
- Display the final regions from each of the models on the resulting map from the statistic. If the regions fall within the output from this statistic, the important aspects of the models are in agreement.
- Examine if the output values are clustered or if they are fragmented. If they are fragmented and the final regions were created, run the Region Overlap statistics to examine if the regions are in agreement.
Percent Change in High Suitability
See Percent Change in High Suitability for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for:
- Examine whether the positive and negative values are separately clustered.
- If you are exploring a change in a weight, transformation, or criteria, can you see the effects of the change from one model to another?
- See where positive values are since they indicate where Model 1 has the higher suitability values, and negative values indicate that the change produces more favorable results.
Models Similar and Different with High Suitability
See Models Similar and Different with High Suitability for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for:
- Observe the locations that differ between the models that have high suitability. You may need to use the output from the Models Similar in High Suitability statistic in conjunction with the output from this statistic to determine these locations. These locations pinpoint areas of disagreement in the models highlighting the most significant discrepancies between them.
- Ensure that the final regions fall within the locations where the models are similar and have high suitability.
- Determine the spatial relationship between where the models agree and disagree in high suitability. If the locations are near one another, can you explain why?
- Note that if the differing locations are spread across the study area, a categorical criterion, such as land use, could be influencing the results.
Models Similar with Low Suitability
See Models Similar with Low Suitability for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for:
- Ensure that many locations are similar with low suitability. This indicates your models agree on the areas that are not favorable.
- Increase the Low Suitability threshold parameter and see where the additional locations are added. Do the new locations expand out from the existing locations?
Investigate Change Between Models
There are two statistics in the Investigate Change Between Models functional group that can be used to compare the input models.
Spatial Association between Suitability Values
See Spatial Association between Suitability Values for a description, when to use, and the general formula for the statistic.
What to look for:
- Note that High to High and Low to Low categories are where the models agree.
- Note that Low to High or High to Low are locations where the models disagree. Low to High indicates Model 1 has the lower suitability values relative to Model 2. High to Low indicates the changes made in Model 2 decreased the suitability.
- Examine the spatial patterns in the categories. Are the categories clustering? How do the categories spatially relate to one another?
- In scenario testing, directly examine the effects of changing the weights, transformations, or input criteria in Model 1 has on Model 2.
- Use this statistic in conjunction with the Change in Suitability Values (Change Detection) statistic.
Change in Suitability Values (Change Detection)
See Change in Suitability Values (Change Detection) for a description, when to use, and the general formula for the statistic.
What to look for:
- For scenario testing, see how changing the weights, transformations, or input criteria in Model 1 directly affect Model 2.
- Analyze the patterns in the changes you made in Model 1. Try to explain how the changes in Model 1 are realized in Model 2.
Examine Where Regions Overlap
There are two statistics in the Examine Where Regions Overlap functional group that can be used to compare the input models.
Region Overlap
See Region Overlap for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for:
- A lot of overlap in the regions indicates that even if the criteria, weights, and transformations might differ, the models result in similar regions.
- If there are multiple regions, examine if there are any patterns among the regions. Examine if some of the regions overlap a great deal and others not. Can you relate any of these patterns back to the original criteria and transformations?
- Note that if there is no overlap in the regions, the models are independent of one another.
Region Overlap in Suitability Levels
See Region Overlap in Suitability Levels for a description of the statistic, when to use it, and the general formula underlying the statistic.
What to look for:
- Examine where the regions do not overlap. Are they in areas with high suitability? If so, even though the regions do not exactly align, the resulting regions are still favorable since they are in high suitability.
- Identify the locations that overlap or not that fall in low suitability. These are the locations to be most concerned with. At these locations, the models differ significantly, reducing the mean of their suitability values.
- See what to look for in the Region Overlap statistic description above for additional insights.