Maps

Output from this procedure also includes interactive maps.

  1. Double-click the first output object that displays a horizontal bar chart and a map to activate it.
    Figure 1. Bar chart and map of rules ranked by confidence
    Bar chart and map of rules ranked by confidence
    • The bar chart displays rules ranked by confidence. Confidence is the percentage of correct rule predictions: the percentage of cases for which the predicted value is correct when the conditions of the rule are true.
    • The blue points on the map are events that satisfy the rule. Both the conditions and the predicted outcome are true for those events.
    • The darker green regions are regions that meet the conditions of the rule. The overall value of each condition of the rule is true for that region.
    • The red points on the map are events that have the predicted value but do not meet all the conditions of the rule.
    • The lighter green regions of the map are regions that do not meet one or more of the conditions of the rule.
  2. To see the map results for a different rule, click the bar for that rule. For example, click rule 5.
    Figure 2. Rule 5 ranked by confidence
    Rule 5 ranked by confidence
    • This rule predicts a crime type of minor theft in regions where the average property value is below a certain value.
    • Regions that meet the conditions of the rule tend to be in the central area of the map.
  3. Click rule 8 in the bar chart.
    Figure 3. Rule 8 ranked by confidence
    Rule 8 ranked by confidence
    • This rule predicts a crime time of major theft in regions where the population is greater than a certain value and the average property value is above a certain value.
    • Regions that meet the conditions of this rule tend to be on the outer perimeter of the map.
  4. You can also change the measure that is used to determine the relative ranks of the rules in the bar chart. For example, select Rule Support from the Sort list. Then, click on rule 16.
    Figure 4. Rules ranked by rule support
    Rules ranked by rule support
    • Rule support is the percentage of cases for which the rule is true. For example, if the rule is "If property value is less than x and population is greater than y, then type of crime is z," rule support is the actual percentage of census tracts for which the property value is less than x, the population is greater than y, and the most common (modal) crime is z.
    • Rule 16 predicts violent crime in regions where the average property value is less than a certain value.
    • Regions that meet the conditions of the rule tend to be in the central area of the map.

Next