Systems and methods for using spatial and temporal analysis to associate data sources with mobile devices
Various embodiments of the present technology generally relate to data delivery. More specifically, some embodiments of the present technology relate to systems and methods for using spatial and temporal analysis to associate data sources with mobile devices. The delivery of data to support a wide variety of services for and about mobile devices that are based on data stored in corporate, commercial, and government databases which is not currently linked to individual mobile devices. Some embodiments allow advertisers to better target their ads to relevant target audience with greater accuracy.
1. A method of estimating a household location of a user from signals transmitted by a mobile device owned by the user, the mobile device and the household location each having a unique identifier, method comprising:
receiving a multiplicity of data signals from the mobile device over time;
modeling the data signals into at least one time-series of latitude-longitude pairs;
clustering each of the latitude-longitude pairs into one of a number of clusters corresponding to geographic regions visited by the mobile device,
wherein the clustering uses a density-based scan algorithm to find a central location of groups of the latitude-longitude pairs;
calculating a score for each cluster,
wherein calculating a score for each cluster includes basing the score, at least in part, on at least one of a number of location measurements, a local time of occurrence of each location measurement, and a proximity of a centroid of the cluster to the household location; and
wherein the location measurements refer to advertisement requests that include taking into account each application or Website from which the advertisement requests are obtained; and
identifying the user's household location by comparing the cluster scores.
2. The method of claim 1 , wherein the data signals are associated with a time series of device locations.
3. The method of claim 2 , wherein the data signals are associated with a time series of advertisement requests.
4. The method of claim 1 , wherein the data signals comprise a latitude-longitude 2-tuple.
5. The method of claim 4 , wherein data signals further comprise a timestamp at which a location measurement is taken.
6. The method of claim 1 further comprising performing outlier rejection by comparing the at least one time-series of latitude-longitude pairs against a set of physical constraints.
7. The method of claim 1 , wherein clustering the latitude-longitude pairs into a number of clusters is based on geographical proximity over time.
8. The method of claim 1 further comprising associating the mobile device's unique identifier with the unique identifier of the identified user's household location.
9. The method of claim 1 wherein calculating a score for each cluster further comprises taking into account at least one of a total number of location measurements for a pre-determined period of time and a likelihood that the cluster corresponds to the user's household location for the pre-determined period of time.
10. A system for estimating a household location of a user from signals transmitted by a mobile device owned by the user, the mobile device and the household location each having a unique identifier, the system comprising:
a geo-cluster unit structured and arranged to receive a multiplicity of data signals from the mobile device over time, model the data signals into at least one time-series of latitude-longitude pairs, and to cluster each of the latitude-longitude pairs into one of a number of clusters corresponding to geographic regions visited by the mobile device,
wherein to cluster each of the latitude-longitude pairs into one of the number of clusters, the geo-cluster unit uses a density-based scan algorithm to find a central location of groups of the latitude-longitude pairs;
a cluster scoring unit structured and arranged to calculate a score for each cluster,
wherein the cluster scoring unit is structured and arranged to base the score, at least in part, on at least one of a number of location measurements, a local time of occurrence of each location measurement, and a proximity of a centroid of the cluster to the household location; and
wherein the location measurements refer to advertisement requests that include taking into account each application or Website from which the advertisement requests are obtained; and
a cluster selection unit structured and arranged to identify the user's household location by comparing the cluster scores.
11. The system of claim 10 further comprising a statistical scoring unit structured and arranged to account for at least one of a total number of location measurements for a pre-determined period of time and a likelihood that the cluster corresponds to the user's household location for the pre-determined period of time.
12. The system of claim 10 further comprising a latitude-longitude outlier rejection unit structured and arranged to perform outlier rejection by checking the at least one time-series of latitude-longitude pairs against a set of physical constraints.
13. A computer program product embodied on a non-transitory computer-readable medium and comprising computer code for estimating a household location of a user from signals transmitted by a mobile device owned by the user, the mobile device and the household location each having a unique identifier, the code comprising instructions stored in a data storage medium and executable on a processing device; and, when executed by the processing device, the processing device:
receives a multiplicity of data signals from the mobile device over time; models the data signals into at least one time-series of latitude-longitude pairs;
clusters each of the latitude-longitude pairs into one of a number of clusters corresponding to geographic regions visited by the mobile device,
wherein to cluster each of the latitude-longitude pairs into one of the number of clusters, the instructions cause the processing device to use a density-based scan algorithm to find a central location of groups of the latitude-longitude pairs;
calculates a score for each cluster,
wherein to calculate the score for each cluster includes basing the score, at least in part, on at least one of a number of location measurements, a local time of occurrence of each location measurement, and a proximity of a centroid of the cluster to the household location; and
wherein the location measurements refers to advertisement requests that include taking into account each application or Website from which the advertisement request is obtained; and
identifies the user's household location by comparing the cluster scores.
14. The computer program product claim 13 further comprising code comprising instructions for performing outlier rejection by checking the at least one time-series of latitude-longitude pairs against a set of physical constraints.
15. The computer program product of claim 13 wherein clustering the latitude-longitude pairs into a number of clusters is based on geographical proximity over time.
16. The computer program product of claim 13 further comprising code comprising instructions for associating the mobile device's unique identifier with the unique identifier of the identified user's household location.
17. The computer program product of claim 13 , wherein code instructions for calculating a score for each cluster further include at least one of accounting for a total number of location measurements for a pre-determined period of time and a likelihood that the cluster corresponds to the user's household location for the pre-determined period of time.