Geographic locale mapping system for outcome prediction
When applied to geographical locale datasets, cluster characterization and comparison techniques meaningfully transform representations of data to facilitate sophisticated locale-specific comparisons and provide location-specific outcome predictions. Outcome predictions may, for example, be based on distributions of locale attributes associated with of a large number of geographic locales and/or by computing statistical probabilities of particular outcomes based on outcome data associated with each of the geographic locales.
1. A system for geographic locale outcome prediction, the system comprising:
store within a database cluster data for a plurality of clusters each representing a different geographic locale and defined by a common set of locale attributes;
a prediction module stored in memory and executable by a processor, the prediction module configured to:
access the cluster data for the plurality of clusters, each one of the plurality of clusters being classified in the database by a select one of multiple predefined cluster types, the multiple predefined cluster types being derived in a clustering analysis of the locale attributes for the plurality of clusters and each classifying a subgroup of the multiple clusters having locale attributes with a demonstrated correlation to one another;
receive an input including information for identifying a benchmark cluster corresponding to a geographic locale where an outcome of interest has been observed, the benchmark cluster being classified in the database by a benchmark cluster type that is one of the multiple predefined cluster types;
query the database with the benchmark cluster type to identify one or more other geographic locales associated in the database with the benchmark cluster type; and
output the one or more identified geographic locales, the one or more identified geographic locales representing locales where the outcome of interest is predicted to occur.
2. The system of claim 1 , wherein receiving the input includes receiving an address.
3. The system of claim 1 , wherein the multiple predefined cluster types are defined based on an assessment of one or more predictive locale attributes identified among the common set of locale attributes, each of the one or more predictive locale attributes characterized by a distribution having a standard deviation in excess of a predetermined threshold.
4. The system of claim 1 , wherein the cluster data further includes outcome data for the specified type of event in association with a statistically significant number of the different geographic locales and the processor assesses the similarity metric by quantifying a predictive likelihood of the outcome of interest occurring in each of the plurality of clusters based on the outcome data.
5. One or more non-transitory computer-readable storage media of a tangible article of manufacture encoding computer-executable instructions for executing on a computer system a computer process, the computer process comprising:
receiving an input including information for identifying a benchmark cluster in a database corresponding to a geographic locale where an outcome of interest has been observed, the benchmark cluster being one of a plurality of clusters defined by a set of locale attributes and associated in the database with a select one of multiple predefined cluster types, the multiple predefined cluster types being derived in a clustering analysis of the locale attributes for the plurality of clusters such that each of the multiple predefined cluster types classifies a subgroup of the plurality of clusters having locale attributes with a demonstrated correlation to one another;
determining for the benchmark cluster a benchmark cluster type that is one of the multiple predefined cluster types;
querying the database with the benchmark cluster type to identify one or more other geographic locales associated in the database with the benchmark cluster type; and
outputting the one or more identified geographic locales, the one or more identified geographic locales representing locales where the outcome of interest is predicted to occur.
6. The one or more non-transitory computer-readable storage media of claim 5 , wherein receiving the input includes receiving an address.
7. The one or more non-transitory computer-readable storage media of claim 5 , further comprising:
for each locale attribute of the common set of locale attributes, assessing a distribution of corresponding values in the dataset to identify at least one predictive locale attribute characterized by a distribution having a standard deviation in excess of a predetermined threshold; and
analyzing one or more distributions of values for the at least one predictive locale attribute to define a number of different cluster types each representing one of the multiple predefined cluster types.
8. The one or more non-transitory computer-readable storage media of claim 7 , wherein analyzing the one or more distributions to define a number of different cluster types further comprises:
computing, for each cluster, a sum of square distances, each of the square distances representing a distance between a centroid of the cluster and one of the locale attributes;
plotting a curve representing the computed sum of square distances for each cluster;
identifying local extrema on the plotted curve; and
defining each local extrema as a different cluster type.
9. The one or more non-transitory computer-readable storage media of claim 8 , further comprising:
classifying each of the plurality of clusters as one of the multiple predefined cluster types.
10. The one or more non-transitory computer-readable storage media of claim 5 , wherein the cluster data further includes outcome data for the specified type of event in association with a statistically significant number of the different geographic locales and identifying the one or more other geographic locales further comprises quantifying a predictive likelihood of the outcome of interest occurring in each of the plurality of clusters based on the outcome data.
11. One or more non-transitory computer-readable storage media of a tangible article of manufacture encoding computer-executable instructions for executing on a computer system a computer process, the computer process comprising:
receiving an input including information for identifying a benchmark cluster in a database corresponding to a geographic locale associated with an observed outcome of interest, the benchmark cluster being one of a plurality of clusters defined by a set of locale attributes and associated in the database with a select one of multiple predefined cluster types, the multiple predefined cluster types being derived in a clustering analysis of the locale attributes for the plurality of clusters such that each of the multiple predefined cluster types classifies a subgroup of the plurality of clusters having locale attributes with a demonstrated correlation to one another;
determining for the benchmark cluster a benchmark cluster type that is one of the multiple predefined cluster types;
querying the database with the benchmark cluster type to identify one or more geographic locales associated in the database in association with the benchmark cluster type; and
outputting the one or more identified geographic locales, the one or more identified geographic locales representing locales where the outcome of interest is predicted to occur.