RISK MAPPING
Inverse Distance Weighting (IDW) is a mapping technique that is useful for highlighting local patterns. We have used IDW in the Risk Mapping maps in order to highlight local areas with a concentration of high-risk postcodes. Mapping Risk Mapping indices is not unproblematic; below-average values are constrained between 0 and 100, but there is no upper limit for above-average values. This creates problems for ranges and class intervals with some very high values (Harris et al. 2005). Logarithmic transformation of the indices results in a more symmetric distribution.


The Risk Mapping maps are based on percentiles from the distribution of the log transformed values: Very low (0-25%), Low (26-50%), Medium (51-75%), High (76-95%) and Very High (95+%) (Table).
Risk index classes based on percentiles in log-transformed values of the Mosaic Grand Index. [TOP]
| Title | Index | Percentile | Log Index |
| Very Low | 1-69 | 0-25% | 0-1.838 |
| Low | 70-99 | 26-50% | 1.839-1.999 |
| Medium | 100-123 | 51-75% | 2.000-2.090 |
| High | 124-217 | 76-95% | 2.092-2.337 |
| Very High | 218+ | 95+% | 2.338+ |
Log-transforming has two effects on the Risk Mapping surfaces:
- it increases the specificity and spatial accuracy of the High and Very High risk classes;
- it decreases the specificity of the Low and Very Low risk classes.
Targeting areas on the basis of risk prediction is mostly concerned with High and Very High risk classes and the log-transformed indices are for that reason advantageous relative to the raw index values. However, there are also cases where the 'cold spots' would form the main policy focus and the un-transformed values would result in greater specificity at this end of the scale.
REFERENCES [TOP]
Harris, R. et al. 2005. Geodemographics, GIS and neighbourhood targeting. Chichester: John Wiley & Sons Ltd.


