Machine-learning model for detecting a device within a venue
A model is configured to determine whether a device is located within a venue. During a baseline time period, the system detects wireless pings from mobile devices. The system obtains device parameters from the wireless pings. The system evaluates the device parameters to determine whether a mobile device entered the venue or remained outside of the venue. The system trains a model on training data corresponding to the baseline time period, the model configured to differentiate between devices that enter the venue and devices that remain outside the venue based on device parameters associated with the device. The system applies the model to future detected devices to determine whether or not the devices enter the venue.
1. A non-transitory computer readable storage medium comprising stored instructions, the instructions when executed cause at least one processor to:
generate a set of training data representative of devices previously located within boundaries associated with a physical structure and devices previously located outside the boundaries associated with the physical structure, the set of training data including characteristics of pings received from the devices previously located within the boundaries and outside the boundaries and including labels indicative of whether a device each ping was received from was located within the boundaries or outside the boundaries, wherein the pings received at times outside of hours of operation of the physical structure are labeled as located outside the boundaries;
train a neural network specific to the physical structure using the generated set of training data, wherein the neural network is configured to determine, based on weighted scores assigned to device parameters of a mobile device, whether the mobile device is physically located within the boundaries associated with the physical structure;
detect, via a plurality of wireless access points of the physical structure, a plurality of pings from a device;
measure, by each wireless access point, a signal strength associated with each ping detected by the wireless access point; and
determine whether the device is physically located within the boundaries associated with the physical structure by applying the neural network to the signal strengths associated with the plurality of pings measured by the plurality of wireless access points.
2. The non-transitory computer readable storage medium of claim 1 , wherein the device parameters comprise at least one of: a signal strength of a ping, a time of the ping, a dwell time between a first ping and a last ping from a device, whether the ping was received during hours of operation of the physical structure, a signal strength of pings from other devices that connected to the wireless access point, and a manufacturer identifier of a MAC address of the device.
3. The non-transitory computer readable storage medium of claim 1 , wherein the neural network is further configured to differentiate between devices located within the boundaries associated with the physical structure and outside the boundaries associated with the physical structure based on device parameter values associated with the devices.
4. The non-transitory computer readable storage medium of claim 1 , wherein the neural network is configured to produce a confidence score representative of a likelihood that the device is located within the boundaries associated with the physical structure.
5. The non-transitory computer readable storage medium of claim 1 , wherein the instructions when executed cause the at least one processor to:
identify a media access control (MAC) address of the device; and
determine, based on the MAC address, that a user of the device had previously viewed an advertisement for the physical structure.
6. A method comprising:
generating a set of training data representative of devices previously located within boundaries associated with a physical structure and device previously located outside the boundaries associated with the physical structure, the set of training data including characteristics of pings received from the devices previously located within the boundaries and outside the boundaries and including labels indicative of whether a device each ping was received from was located within the boundaries or outside the boundaries, wherein the pings received at times outside of hours of operation of the physical structure are labeled as located outside the boundaries;
training a neural network specific to the physical structure using the generated set of training data, wherein the neural network is configured to determine, based on weighted scores assigned to device parameters of a mobile device, whether the mobile device is physically located within the boundaries associated with the physical structure;
detecting, via a plurality of wireless access point of the physical structure, a plurality of pings from a device;
measuring, by each wireless access point, a signal strength associated with each ping detected by the wireless access point; and
determining whether the device is physically located within the boundaries associated with the physical structure by applying the neural network to the signal strengths associated with the plurality of pings measured by the plurality of wireless access points.
7. The method of claim 6 , wherein the device parameters comprise at least one of: a signal strength of a ping, a time of the ping, a dwell time between a first ping and a last ping from a device, whether the ping was received during hours of operation of the physical structure, a signal strength of pings from other devices that connected to the wireless access point, and a manufacturer identifier of a MAC address of the device.
8. The method of claim 6 , wherein the neural network is further configured to differentiate between devices located within the boundaries associated with the physical structure and outside the boundaries associated with the physical structure based on device parameter values associated with the devices.
9. The method of claim 6 , wherein the neural network is configured to produce a confidence score representative of a likelihood that the device is located within the boundaries associated with the physical structure.
10. The method of claim 6 , further comprising:
identifying a media access control (MAC) address of the device; and
determining, based on the MAC address, that a user of the device had previously viewed an advertisement for the physical structure.