Detecting a user's outlier days using data sensed by the user's electronic devices
A method for detecting a user's outlier days uses data corresponding to features of the user acquired over multiple days by sensors on the user's electronic device. The data acquired for each day and feature is labeled as regular or irregular by applying N labeling approaches. One of the N labeling approaches compares the data for each feature with how values of previously acquired data for corresponding features are distributed. N labels are generated for the data for each feature and day. The machine learning classification model is trained using one of the N labels for each of the N labeling approaches. An optimal labeling approach is selected from among the N labeling approaches for each feature using the machine learning classification model. For each feature, the method determines whether each of the days is an outlier day for the user using the labels obtained with the optimal labeling approach.
1 . A method for detecting an outlier time period of a user, comprising:
receiving data corresponding to each of a plurality of features related to the user, wherein the data is acquired over a plurality of time periods by sensors on an electronic device of the user;
labeling the data acquired on each of the plurality of time periods for each of the plurality of features as being regular or irregular by applying N labeling approaches, wherein N is a positive integer greater than one, and wherein at least one of the N labeling approaches involves comparing the data corresponding to each of the plurality of features with how values of previously acquired data for corresponding features are distributed;
generating N labels as synthetic features for the data corresponding to each of the plurality of features for each of the plurality of time periods;
training a machine learning classification model based at least in part on (i) analyzing the synthetic features labeled as irregular for each of the N labeling approaches and (ii) determining a regularization parameter using a grid search; and
using the machine learning classification model:
selecting from among the N labeling approaches an optimal labeling approach for each of the plurality of features; and
determining for each of the plurality of features, using the labels obtained with the optimal labeling approach and by assigning a relevancy to each of the plurality of features, whether each of the plurality of time periods is an outlier time period of the user.
2 . The method of claim 1 , wherein each of the plurality of time periods is selected from the group consisting of: a day, a work day, a 10-hour period, a week, a month, and a year.
3 . The method of claim 1 , wherein the plurality of features includes a feature selected from the group consisting of: a usage pattern of the electronic device by the user, location data indicative of a location of the user, environmental data related to the location of the user, physiological data of the user, and behavioral data of the user.
4 . The method of claim 3 , wherein the location data relates to a parameter selected from the group consisting of: indoor temperature, indoor light level, light exposure, and noise level, wherein the environmental data relates to a parameter selected from the group consisting of: pollution level, outdoor light level, humidity, and outdoor temperature, wherein the physiological data relates to a parameter selected from the group consisting of: galvanic skin response, heart rate variability, and skin temperature, and wherein the behavioral data relates to a parameter selected from the group consisting of: mobility, social interactions, and sleep state.
5 . The method of claim 1 , wherein one of the N labeling approaches involves comparing the data corresponding to each of the plurality of features with an i and 100-i percentile of a distribution of the previously acquired data for features corresponding to each of the plurality of features, and wherein i is a design parameter.
6 . The method of claim 1 , wherein one of the N labeling approaches involves comparing the data corresponding to each of the plurality of features with a mean of a distribution of the previously acquired data for features corresponding to each of the plurality of features.
7 . The method of claim 1 , wherein one of the N labeling approaches involves comparing the data corresponding to each of the plurality of features with a variance of a distribution of the previously acquired data for features corresponding to each of the plurality of features.
8 . The method of claim 1 , further comprising:
determining for each of the plurality of features whether each of the plurality of time periods is a positive outlier time period or a negative outlier time period for the user.
9 . The method of claim 1 , wherein the plurality of features related to the user includes a wellbeing feature of the user.
10 . The method of claim 1 , further comprising:
generating a recommendation for the user based on the relevancy of each of the plurality of features.
11 . The method of claim 1 , wherein the selecting the optimal labeling approach for each of the plurality of features is performed in an electronic processing device remote from the user, and wherein the data that is acquired by sensors on the electronic device of the user is transmitted through a communications network to the remote processing device.
12 . The method of claim 1 , wherein the sensors on the electronic device of the user are selected from the group consisting of: a motion sensor, a location sensor, a temperature sensor, an air quality sensor, a pedometer, a light sensor, a battery level sensor, and an ambient noise level sensor.
13 . The method of claim 1 , wherein the electronic device of the user is selected from the group consisting of: a mobile phone, a smartwatch, a smartphone, a laptop, a tablet computer, and a personal computer.
14 . A system for detecting an outlier time period of a user, comprising:
a receiver that receives data through a communications network from an electronic device of the user, wherein the data corresponds to each of a plurality of features related to the user, and wherein the data is acquired during a plurality of time periods by sensors on the electronic device of the user; and
a computer including a processor, wherein the processor is configured to:
label the data acquired during each of the plurality of time periods for each of the plurality of features as being regular or irregular by applying N labeling approaches, wherein N is a positive integer greater than one, and wherein at least one of the N labeling approaches involves comparing the data corresponding to each of the plurality of features with how values of previously acquired data for corresponding features are distributed;
generate N labels as synthetic features for the data corresponding to each of the plurality of features for each of the plurality of time periods;
train a machine learning classification model based at least in part on (i) analyzing the synthetic features labeled as irregular for each of the N labeling approaches and (ii) determining a regularization parameter using a grid search; and
using the machine learning classification model:
select from among the N labeling approaches an optimal labeling approach for each of the plurality of features using the machine learning classification model; and
determine for each of the plurality of features, using the labels obtained with the optimal labeling approach and by assigning a relevancy to each of the plurality of features, whether each of the plurality of time periods is an outlier time period of the user.
15 . The system of claim 14 , wherein the plurality of features includes a feature selected from the group consisting of: a usage pattern of the electronic device by the user, location data indicative of a location of the user, environmental data related to the location of the user, physiological data of the user, and behavioral data of the user.
16 . The system of claim 15 , wherein the location data relates to a parameter selected from the group consisting of: indoor temperature, indoor light level, light exposure, and noise level, wherein the environmental data relates to a parameter selected from the group consisting of: pollution level, outdoor light level, humidity, and outdoor temperature, wherein the physiological data relates to a parameter selected from the group consisting of: galvanic skin response, heart rate variability, and skin temperature, and wherein the behavioral data relates to a parameter selected from the group consisting of: mobility, social interactions, and sleep state.
17 . The system of claim 14 , wherein one of the N labeling approaches involves comparing the data corresponding to each of the plurality of features with a mean of a distribution of the previously acquired data for features corresponding to each of the plurality of features.
18 . The system of claim 14 , wherein the sensors on the electronic device of the user are selected from the group consisting of: a motion sensor, a location sensor, a temperature sensor, an air quality sensor, a pedometer, a light sensor, a battery level sensor, and an ambient noise level sensor.
19 . The system of claim 14 , wherein the electronic device of the user is selected from the group consisting of: a mobile phone, a smartwatch, a smartphone, a laptop, a tablet computer, and a personal computer.