Computer model accuracy in predicting location of online system users
An online system receives a request from a user of a manager transmitter to generate a unique beacon identifier (ID) associated with a physical location. Responsive to receiving the beacon ID from the online system, the manager transmitter transmits a Bluetooth signal comprising the beacon ID to user client devices, which send the beacon ID to the online system for identification. Responsive to detecting that a received signal strength exceeds a threshold, a location context module classifies the instance of the user client device detecting the signal as an example of a user being present at the physical location. A location prediction module uses the instance as training data to train a machine-learning model to predict the presence of online system users at the physical location.
1. A method comprising:
receiving, by an online system, from a first client device associated with a first user of an online system, a request to identify a short-range radio signal;
identifying, by the online system, the short-range radio signal as associated with a physical location;
responsive to detecting that a signal strength associated with the short-range radio signal exceeds a threshold, classifying, by the online system, the first client device as present at the physical location;
receiving, by the online system, a first set of context information comprising signals received by the first client device when present at the physical location; and
training, by the online system, a machine-learning model using training data including the first set of context information to predict a likelihood that a second user of the online system is present at the physical location based on a second set of context information comprising signals received from a second client device associated with the second user.
2. The method of claim 1 , further comprising:
verifying, by the online system, a position of a transmitter associated with the short-range radio signal; and
responsive to the transmitter being located within a threshold distance of the physical location, adding, by the online system, the first set of context information to the training data.
3. The method of claim 1 , wherein the first or second set of context information comprises one or more of: visible wireless networks, visible Bluetooth signals, altimeter reading, and time of day.
4. The method of claim 1 , wherein the training data further comprises instances of client devices not being present at the physical location.
5. The method of claim 1 , further comprising targeting, by the online system, delivery of content items to the second client device associated with the second user responsive to the prediction.
6. The method of claim 1 , further comprising instructing, by the online system, a transmitter associated with the short-range radio signal to transmit the short-range radio signal responsive to detecting that the transmitter is present at the physical location.
7. The method of claim 1 , further comprising instructing, by the online system, a transmitter associated with the short-range radio signal to transmit the short-range radio signal responsive to user input comprising a request to begin transmission.
8. The method of claim 1 , further comprising applying, by the online system, a trained validation model to quantify accuracy of the machine-learning model.
9. The method of claim 1 , further comprising iteratively re-training, by the online system, the machine-learning model responsive to the machine-learning model not reaching an accuracy measurement over a threshold.
10. The method of claim 1 , further comprising weighting, by the online system, a value of the training data based on the signal strength of the short-range radio signal.
11. A non-transitory computer readable storage medium storing instructions that, when executed by a computing system, cause the computing system to perform operations comprising:
receiving, by an online system, from a first client device associated with a first user of an online system, a request to identify a short-range radio signal;
identifying, by the online system, the short-range radio signal as associated with a physical location;
responsive to detecting that a signal strength associated with the short-range radio signal exceeds a threshold, classifying, by the online system, the first client device as present at the physical location;
receiving, by the online system, a first set of context information comprising signals received by the first client device when present at the physical location; and
training, by the online system, a machine-learning model using training data including the first set of context information to predict a likelihood that a second user of the online system is present at the physical location based on a second set of context information comprising signals received from a second client device associated with the second user.
12. The non-transitory computer readable storage medium of claim 11 , further comprising:
verifying, by the online system, a position of a transmitter associated with the short-range radio signal; and
responsive to the transmitter being located within a threshold distance of the physical location, adding, by the online system, the first set of context information to the training data.
13. The non-transitory computer readable storage medium of claim 11 , wherein the first or second set of context information comprises one or more of: visible wireless networks, visible Bluetooth signals, altimeter reading, and time of day.
14. The non-transitory computer readable storage medium of claim 11 , wherein the training data further comprises instances of client devices not being present at the physical location.
15. The non-transitory computer readable storage medium of claim 11 , further comprising targeting, by the online system, delivery of content items to the second client device associated with the second user responsive to the prediction.
16. The non-transitory computer readable storage medium of claim 11 , further comprising instructing, by the online system, a transmitter associated with the short-range radio signal to transmit the short-range radio signal responsive to detecting that the transmitter is present at the physical location.
17. The non-transitory computer readable storage medium of claim 11 , further comprising instructing, by the online system, a transmitter associated with the short-range radio signal to transmit the short-range radio signal responsive to user input comprising a request to begin transmission.
18. The non-transitory computer readable storage medium of claim 11 , further comprising applying, by the online system, a trained validation model to quantify accuracy of the machine-learning model.
19. The non-transitory computer readable storage medium of claim 11 , further comprising iteratively re-training, by the online system, the machine-learning model responsive to the machine-learning module not reaching an accuracy measurement over a threshold.
20. The non-transitory computer readable storage medium of claim 11 , further comprising weighting, by the online system, a value of the training data based on the signal strength of the short-range radio signal.