Sensor vectors based on activations of sensors with attributes

A computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:

• calculate a first number of activations of a first set of sensors corresponding to a first sensor attribute, wherein the first number of activations takes place in a time window;
• calculate a second number of activations of a second set of sensors corresponding to a second sensor attribute, wherein the second number of activations take place in the time window, wherein the first set of sensors overlaps the second set of sensors, and wherein the first set of sensors and the second set of sensors monitor a location;
• generate a sensor vector that includes the first sensor attribute with a first assigned value equal to the first number of activations of the first set of sensors and the second sensor attribute with a second assigned value equal to the second number of activations of the second set of sensors;
• and
• identify a pattern based on the sensor vector including the first number of activations and the second number of activations, wherein the pattern corresponds to an activity taking place at the location.

A method comprising:

• converting sensor data from sensors each associated with multiple sensor categories into a sensor vector including data spanning a time window, wherein the sensor vector includes a count of sensor activations for a first sensor category of the multiple sensor categories and a count of sensor activations for a second sensor category of the multiple sensor categories;
• mapping the sensor vector to sensor points in a multi-dimensional latent space;
• and
• determining clusters of the sensor points in the multi-dimensional latent space based on a distance between the sensor points in the multi-dimensional latent space.