Method and apparatus for detecting state

A method for detecting a state, comprising:

• receiving a classifier sent by a classifier generating device, wherein the classifier is acquired by training a state sample set determined as training data by the classifier generating device according to a classification algorithm and is configured to determine whether a mobile terminal to be detected is located in a holding space similar to a preset holding space, the state sample set comprises a first plurality of holding state samples collected when a sample mobile terminal is located in the preset holding space, and a second plurality of non-holding state samples collected when the sample mobile terminal is not located in the preset holding space, wherein a missing value supplementation processing is performed on the first plurality of holding state samples and the second plurality of non-holding state samples by using a mean value of a number of feature parameters in a same category, and a format conversion is performed on the classifier to enable the classifier to be suitable to an operation environment of a target mobile terminal, wherein the format conversion on the classifier comprises converting a binary file of the classifier in a spark format into a pmml file of the classifier in a pmml format;
• acquiring a current state sample of the mobile terminal to be detected at a current time;
• inputting the current state sample into the classifier;
• and
• acquiring a label outputted by the classifier according to the current state sample of the mobile terminal to be detected at the current time, wherein the label is configured to indicate a category of the current state sample, and the category of the state sample is one of a category of locating in the holding space and a category of not locating in the holding space.

An apparatus for detecting a state of a target mobile terminal, comprising at least one processor configured to:

• acquire a state sample set of a plurality of sample mobile terminals, wherein the state sample set comprises a first plurality of holding state samples collected by the plurality of sample mobile terminals when the plurality of sample mobile terminals are located in a preset holding space, and a second plurality of non-holding state samples collected by the plurality of sample mobile terminals when the plurality of sample mobile terminals are not located in the preset holding space, and a missing value supplementation processing is performed on the first plurality of holding state samples and the second plurality of non-holding state samples by using a mean value of a number of feature parameters in a same category;
• determine a classification algorithm;
• acquire a classifier by training the state sample set determined as training data according to the classification algorithm;
• send the classifier to the target mobile terminal, in which the target mobile terminal is configured to determine whether the target mobile terminal is located in a holding space similar to the preset holding space based on the classifier and according to a state sample of the target mobile terminal;
• perform a format conversion on the classifier, to enable the classifier to be suitable to an operation environment of the target mobile terminal, wherein performing the format conversion on the classifier comprises converting a binary file of the classifier in a spark format into a pmml file of the classifier in a pmml format;
• and
• send the converted classifier to the target mobile terminal.