JP6545115B2 - LIFE PATTERN PREDICTION DEVICE, PORTABLE TERMINAL, PROGRAM, AND METHOD FOR PREDICTING LIFE PATTERN OF FUTURE PERIOD OF USER - Google Patents

Description

  The present invention relates to a technique for predicting a user's behavior by analyzing a past log in position information of a portable terminal.

  In the communication facility, since the history of the position information of the base station under which the mobile terminal is subordinate is recorded, it is possible to estimate the place of stay and the stay time as the user's action of possessing the mobile terminal using this information. In this case, the mobile terminal does not need to activate the GPS (Global Positioning System) function, nor does it need to transmit positioning information via the network. On the other hand, base station location information has the problem that the spatial granularity is coarse and the time interval is not constant. However, for example, it may be information for determining a rough action of the user such as “stay”.

  2. Description of the Related Art Conventionally, there is a technique for recording an action log that is assumed to have participated in a specific event when the user moves a specific geographical route (see, for example, Patent Document 1). According to this technique, when the user travels the geographical route and can participate in the event, the content regarding the event is provided to the user.

JP, 2014-139837, A

  According to the prior art described above, after the position information of the portable terminal possessed by the user is clearly specified, the information according to the purpose of the stay is provided to the portable terminal.

  On the other hand, the inventors of the present application may not divide daily activities of an unspecified number of people into about several life patterns? I thought. That is, by analyzing the past daily life patterns for a specific user, for example, can it be possible to predict the day's life patterns? I thought. In this case, information required at an appropriate time can be provided to the portable terminal of the user according to the predicted life pattern.

  Therefore, according to the present invention, it is an object of the present invention to provide a life pattern prediction device, a portable terminal, a program and a method capable of predicting a life pattern of a user's future period.

According to the present invention, in the life pattern predicting device for predicting the lifestyle pattern of future periods Yu over THE,
A machine learning engine constructed using as teaching data the time series life pattern types for each unit period collected from an unspecified number of people;
Position information storage means for storing position information of the portable terminal and time series data of day and time;
Action type extraction means for extracting action types based on stay from time series data ;
Life pattern type assignment means for assigning any given life pattern type based on one or more action types for each unit period;
The machine learning engine is characterized by having a life pattern type prediction means for inputting a time series life pattern type in a past period (> unit period) and predicting a life pattern type of a future period as its output .

According to another embodiment of the life pattern prediction device of the present invention,
It is also preferable that the life pattern type assignment means assign a life pattern type based on a combination of a plurality of action types in a unit period.

According to another embodiment of the life pattern prediction device of the present invention,
It is also preferable that the life pattern type assignment means assigns life pattern types based on the transition of a plurality of action types in a unit period.

According to another embodiment of the life pattern prediction device of the present invention,
It is also preferable that the life pattern type prediction means predict the life pattern type of the future period of the user based on the appearance probability of each life pattern type in the past period.

According to another embodiment of the life pattern prediction device of the present invention,
It is also preferable that the life pattern type prediction means calculate the appearance probability for each life pattern type based on the periodicity of a plurality of life pattern types in the past period.

According to another embodiment of the life pattern prediction device of the present invention,
The life pattern type verification means for verifying whether or not the action types extracted by the action type extraction means are consistent with the action types constituting the life pattern types predicted by the life pattern type prediction means after the start of the future period Furthermore, having is also preferable.

According to another embodiment of the life pattern prediction device of the present invention,
The action type extraction unit further extracts start and end times of each action type,
Behavior time prediction for predicting the start / end time of the action type in the future period of the user based on the life pattern type predicted by the life pattern type prediction means, the action type extracted by the action type extraction means and the start / end time thereof It is also preferred to further include means.

According to another embodiment of the life pattern prediction device of the present invention,
It is also preferable that the action time prediction means apply the distribution of the start / end times of the action types in the life pattern types calculated to have the highest appearance probability by the life pattern type prediction means.

According to another embodiment of the life pattern prediction device of the present invention,
The action time prediction means is
For each life pattern type, calculate the distribution of the start / end time of the action type,
It is also preferable to make a prediction based on the probability of occurrence of each life pattern type and the distribution of the start / end time of the action type.

According to another embodiment of the life pattern prediction device of the present invention,
Behavior time verification means for verifying whether the start / end time of the action type extracted by the action type extraction means matches with the start / end time of the action type predicted by the action time prediction means after the start of the future period Furthermore, having is also preferable.

According to another embodiment of the life pattern prediction device of the present invention,
The unit period is the day,
The past period is consecutive days,
The future period is the next day,
In the teacher data of the machine learning engine, the day of the week is associated with the daily time-series life pattern type,
It is also preferable that the life pattern type prediction means inputs the time series life pattern type in the past consecutive days into the machine learning engine and predicts the life pattern type of the next day as its output.

According to the present invention, in the mobile terminal to predict the life pattern of future periods Yu over THE,
A machine learning engine constructed using as teaching data the time series life pattern types for each unit period collected from an unspecified number of people;
Position information storage means for storing position information of the portable terminal and time series data of day and time;
Action type extraction means for extracting action types based on stay from time series data ;
Life pattern type assignment means for assigning any given life pattern type based on one or more action types for each unit period;
The machine learning engine is characterized by having a life pattern type prediction means for inputting a time series life pattern type in a past period (> unit period) and predicting a life pattern type of a future period as its output .

According to the present invention, in a program that causes a computer mounted on a device to predict a life pattern of a user's future period,
A machine learning engine constructed using as teaching data the time series life pattern types for each unit period collected from an unspecified number of people;
Position information storage means for storing position information of the portable terminal and time series data of day and time;
Action type extraction means for extracting action types based on stay from time series data ;
Life pattern type assignment means for assigning any given life pattern type based on one or more action types for each unit period;
It is characterized in that the computer functions as a life pattern type prediction means for inputting time series life pattern types in the past period (> unit period) to the machine learning engine and predicting life pattern types in the future period as its output. .

According to the present invention, in a method of predicting a life pattern of a device for predicting a life pattern of a user's future period,
The device is
A machine learning engine constructed using as teaching data the time series life pattern types for each unit period collected from an unspecified number of people;
A position information storage unit that time accumulated series data of the position information and the date time of the portable terminal has a <br/>,
Device,
A first step of extracting an activity type based on stay from time series data;
A second step of assigning any predetermined life pattern type based on one or more action types for each unit period;
A machine learning engine is characterized by including a third step of inputting a time-series life pattern type in the past period (> unit period) and predicting a life pattern type of a future period as its output .

  According to the life pattern prediction device, the portable terminal, the program and the method of the present invention, it is possible to predict the life pattern of the future period of the user.

It is a spatial external view showing stay of a portable terminal. It is a functional block diagram of the life pattern prediction apparatus in this invention. It is an explanatory view showing processing of a position information storage part and an action type extraction part. It is an explanatory view showing processing of a living pattern type assignment part. It is an explanatory view showing processing of a living pattern type prediction part. It is an explanatory view showing processing of a living pattern type verification part. It is an explanatory view showing processing of an action time forecasting part. It is explanatory drawing showing the process of an action time verification part. It is a system configuration figure of a life pattern prediction device arranged as a communication equipment device.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a spatial appearance view showing a stay of a portable terminal.

  A mobile terminal (for example, a mobile phone or a smartphone) possessed by the user is always under the base station regardless of the position. It is connected to a wide area wireless communication network (mobile phone network). According to FIG. 1, the user is the “stay place” where the address of the home, the whereabouts of the work, and the vicinity of the D station to be visited are. In addition, the user is moving at places other than his / her home, work and around D station.

  In a communication facility that integrates a large number of base stations, base station location information with coarse spatial granularity and irregular time intervals is recorded for each mobile terminal. "Spatial granularity is coarse" means that the geographical distance between location information is relatively long. Further, "the time interval is not constant" means that the recording time interval of the position information is relatively dispersed.

  The base station connected to the wide area wireless communication network records the date and time as a communication history by communicating with the portable terminal 2 located thereunder. The communication history is not limited to the time of a call requiring user operation to the portable terminal, transmission / reception of mail, and browsing of a web page. Also at the time of transmission and reception of data automatically executed by an application installed in the portable terminal, the base station records the communication history from the portable terminal 2 as communication history.

  FIG. 2 is a functional block diagram of the life pattern prediction apparatus in the present invention.

  The life pattern prediction device 1 in the present invention predicts the life pattern of the user's future period. According to FIG. 2, the position information storage unit 101, the teacher data group 102, the action type extraction unit 11, the life pattern type assignment unit 12, the life pattern type prediction unit 13, the machine learning engine 14, and the life pattern It has a type verification unit 15, an action time prediction unit 16, and an action time verification unit 17. The predicted life pattern is input to the application processing unit, and various services are provided to the user according to the life pattern. These functional components are realized by executing a program that causes a computer installed in the device to function. Moreover, according to FIG. 2, the flow of the process using each function structure part can also be understood as a life pattern estimation method of an apparatus.

  FIG. 3 is an explanatory view showing processing of the position information storage unit and the action type extraction unit.

[Position information storage unit 101]
The position information storage unit 101 stores position information of the portable terminal and time-series data of date and time.
According to FIG. 3, latitude and longitude are represented as position information. The latitude and longitude may be position information of a base station that communicates with the mobile terminal.
The time series data of the position information storage unit 101 is referred to from the action type extraction unit 11.

[Action type extraction unit 11]
The action type extraction unit 11 refers to the time-series data of the position information storage unit 101, and extracts an action type based on the stay.
"Stay" means being located within a predetermined distance range for a predetermined time or more. Alternatively, it may be located within one map range which is equally divided on the map for a predetermined time or more.
The “type of behavior” is the range in which the behavior can be estimated, such as “home stay” which is a stay at home, “work stay” which is a stay at work, or “other stay” which is a stay at other places. It is classified by giving meaning to “stay”, and is a component for typifying “life pattern type” described later. In addition, based on "stay", the period which is not "stay" is extracted, it is considered as "moving", meaning may be added and classified within a range in which the action can be estimated, and this is classified as "life pattern type" It may be a component for
The daily activity type extracted by the activity type extraction unit 11 is output to the life pattern type allocation unit 12.

  The behavior type extraction unit 11 extracts a home position, a work position, and the like characterizing the user's behavior from the position information history. First, a plurality of stay points of the user and the stay start date and the stay end date at each stay point are specified. Then, the number of days of stay is calculated for each of the plurality of stay points, the stay point having the largest number of stay days is extracted as a home stay point, and the stay point having the second largest number of stay days is extracted as a workplace stay point.

  Further, as another embodiment, even if the behavior type extraction unit 11 is a machine learning engine and learns a behavior type-attached time zone (home stay / work stay at a time zone, etc.) as teacher data. Good. In this case, the closest action type can be extracted by inputting the time-series data of the position information storage unit 101 into the machine learning engine.

  According to FIG. 3, the daily activity type of the user is specified from the home stay point and the work stay point. According to this user, it is roughly represented that a home stay is from 21:00 to 7:00 and a work stay is from 9:00 to 19:00.

  FIG. 4 is an explanatory view showing processing of the life pattern type assignment unit.

[Life pattern type assignment unit 12]
The living pattern type assigning unit 12 assigns a predetermined living pattern type to each unit period based on one or more action types.
The unit period may be, for example, 24 hours a day, or one day that is not necessarily fixed at 24 hours, such as the sunrise time at a predetermined place to the sunrise time of the next day, or a predetermined period of 24 hours Only the time zone may be extracted daily. In addition, it may be a short term such as 12 hours for half a day, or a long term such as one week, one month, or one year.

"Life pattern type" is typed based on the type, combination, frequency of appearance, appearance order (transition), etc. of "activity type" (home stay / work stay, etc.) in a unit period (for example, 24 hours a day) It is a template. According to FIG. 4, the human life pattern type can be roughly divided into five as follows.
<Life pattern type A (work type)>
When continuous home stay and continuous work stay exist for more than 2 hours and continuous other stay is 2 hours or less Assumed users: working people (regular work, multiple work), students (regular attending school, part-time job) ,
Housewife (part-time regular work)
<Life pattern type B (going out type)>
Continuous home stay and continuous other stay exist for more than 2 hours and continuous work stay is 2 hours or less Assumed users: working people (holiday), students (weekdays + companions), housewives (weekdays)
<Life pattern type C (home type)>
When continuous home stay exists for more than 2 hours, and continuous work stay and continuous other stay is 2 hours or less Assumed users: working people (holiday), students (holiday), housewives (holiday)
<Life pattern type D (working & going type)>
Consecutive home stay, continuous work stay and other continuous stay longer than 2 hours Assumed users: working people (normal work + companion), students (normal school + companion),
Housewife (part-time work)
<Life pattern type Z (Others)>
When stay at home is less than 2 hours Assumed users: working people (business trip, travel), students (travel), housewives (travel)

The living pattern type assignment unit 12 may assign a living pattern type based on the “combination of a plurality of action types” in a unit period (for example, one day).
Also, the living pattern type assigning unit 12 may assign the living pattern types based on “transition of a plurality of action types” in the unit period.
Life pattern types can be classified into various patterns according to the use of the service.

  The living pattern type assigning unit 12 outputs the assigned living pattern to the living pattern type predicting unit 13.

  FIG. 5 is an explanatory view showing processing of the life pattern type prediction unit.

[Life pattern type prediction unit 13]
The life pattern type prediction unit 13 predicts the life pattern type of the user's future period based on the time series life pattern type in the past period (> unit period). Here, the "period" is defined, for example, as follows.
Unit period: day (for example, 1 day (24 hours))
Past period: consecutive days (eg 2 weeks (14 days))
Future period: The next day (for example, the day (the next day) (24 hours))
Specifically, it is possible to predict the “on the day” life pattern type of the user from the “last two weeks” life pattern type.

  In addition, the life pattern type prediction unit 13 may predict the life pattern type of the future period of the user based on the “appearance probability” of each life pattern type in the past period. Here, the appearance probability for each life pattern type may be calculated based on the “periodicity” of a plurality of life pattern types in the past period.

  The life pattern type prediction unit 13 of the present invention inputs time series life pattern types in the past period to the machine learning engine 14, and predicts life pattern types in the future period as its output.

[Machine learning engine 14]
The machine learning engine 14 is constructed as teacher data, which is a time-series life pattern type for each unit period collected from an unspecified number of people.
Approximately 98% of the unspecified number of people are included in the aforementioned life patterns A to D. According to FIG. 5, it is considered that the machine learning engine 14 will learn the life pattern at the following rate.
Life pattern A: 34%
Life pattern B: 28%
Life pattern C: 24%
Life pattern D: 12%
Life pattern Z: 2%

  The teacher data of the machine learning engine 14 is one in which the day of the week is associated with the daily time-series life pattern type. In this case, the life pattern type prediction unit 13 inputs, to the machine learning engine 14, the time series life pattern types on consecutive days with the previous day of the week, and predicts life pattern types according to the next day as the output. can do.

  Specifically, the machine learning engine 14 may be based on a Bayesian network that predicts the life pattern of the next day according to the life pattern of the consecutive days.

  A "Bayesian network" is a graphical network that describes causal relationships between nodes by probability. The network in the present invention refers to a graph structure in which the life pattern type of consecutive days is "node" and the similarity assigned to the arc between the nodes is assigned as "weight". The Bayesian network is a model of probabilistic reasoning in which the inference of causality among multiple nodes is represented by a directed acyclic graph structure, and the relationships of individual variables are represented by conditional probabilities. The directed acyclic graph structure is one in which an arrow is attached to an arc (directed), and the path of the arrow does not circulate a node. By using the Bayesian network, it is possible to quantitatively estimate the future life pattern type as the probability of the uncertain event from the life pattern types of the past several days which have already passed by.

  The life pattern type predicted by the life pattern type prediction unit 13 is output to the application processing unit.

  FIG. 6 is an explanatory view showing a process of the life pattern type verification unit.

[Life pattern type verification unit 15]
The life pattern type verification unit 15 is optional, and after the start of the future period, the behavior type extracted by the behavior type extraction unit 11 from the time series data of the portable terminal of the user of the day is the life pattern It is verified whether it matches with the action type which constitutes the life pattern type predicted by the type prediction unit 13. The success or failure based on the verification is output to the application processing unit.

  According to the living pattern type verification unit, for example, a user who was predicted to be living pattern type A (working type) on the day, and the application processing unit was scheduled to deliver direct mail for returners according to the return time is actually It is assumed that it is detected that the day is a life pattern type C (home type). In that case, the verification result can be output to the application processing unit, and the mail delivery can be stopped suddenly.

FIG. 7 is an explanatory view showing processing of the action time prediction unit.
[Action time prediction unit 16]
The action type extraction unit 11 further extracts start and end times of each action type. At this time, the action time prediction unit 16 determines the user's future period based on the life pattern type predicted by the life pattern type prediction unit 13, the action type extracted by the action type extraction unit 11, and the start / end time thereof. Predict the start and end time of the action type. The predicted start / end time is output to the application processing unit.

Here, the action time prediction unit 16 can also apply the distribution of the start and end times of the action type in the life pattern type calculated to have the highest appearance probability by the life pattern type prediction unit 13.
Further, the action time prediction unit 16 calculates the distribution of the start / end time of the action type for each life pattern type, and predicts based on the probability of the appearance probability for each life pattern type and the distribution of the start / end time of the action type It may be one.

  FIG. 8 is an explanatory view showing a process of the action time verification unit.

[Action time verification unit 17]
The behavior time verification unit 17 predicts, by the behavior time prediction unit 16, the start / end time of the behavior type extracted by the behavior type extraction unit 11 from the time series data of the portable terminal of the user on the day after the start of the future period. It is verified whether it matches with the start / end time of the specified action type. The success or failure based on the verification is output to the application processing unit.

  According to the action time verification unit 17, for example, it is predicted that the user will return home at 21:00 in the living pattern type A (work type) on the day, and the application processing unit is scheduled to deliver direct mail for returners according to the return time. It is assumed that it is detected that the user has returned home at 18:00 although the user is actually life pattern type A (working type). In that case, the verification result can be output to the application processing unit, and the mail delivery scheduled at 21:00 can be cancelled.

  FIG. 9 is a system configuration diagram of a life pattern prediction device arranged as a communication facility device.

  It is also preferable to configure the life pattern prediction device 1 as a communication facility device installed in a wide area wireless communication network (mobile telephone network). In this case, the living pattern prediction apparatus 1 uses the communication interface unit connected to the wide area communication network, the base station position information management unit, and the communication history collection as an optional function in order to store the communication history in the position information storage unit 101. And a position information history generation unit.

The base station position information management unit associates and stores a base station identifier and base station position information.
The communication history collection unit collects the communication history of the date and time and the base station identifier for each mobile terminal 2 from the base station that connects the mobile terminal 2 under the control.
Communication history (terminal identifier, date time, base station identifier)
The position information history generation unit further associates base station position information corresponding to the base station identifier for each communication history using the base station position information management unit. The communication history is output to the position information storage unit 101.

  As described above in detail, according to the life pattern prediction device, the portable terminal, the program and the method of the present invention, it is possible to predict the life pattern of the future period of the user.

  By predicting the life pattern of the future period, the application processing unit can provide effective information in an appropriate time zone for each user. For example, the evening event information can be distributed to a user who is likely to have free time in the evening of the day. In particular, by analyzing the characteristics of the position where the user appears and not acquiring specific position information (latitude and longitude), the life pattern in the future period of the user can be obtained without requiring the privacy information of the user. It can be predicted.

In the embodiment of the present invention described above, although the life pattern of the future period is predicted, it is not necessarily limited to the prediction of the future period. A living pattern of a past or current specific period during which position information could not be acquired may be estimated. In addition, the user may be notified of the predicted life pattern type itself, or may be advised of health guidance based on the predicted life pattern type.
Also, although it has been described that the base station acquires the position information, the position information may be acquired by another system such as a GPS system.

  For the various embodiments of the present invention described above, various modifications, corrections and omissions of the scope of the technical idea and aspect of the present invention can be easily made by those skilled in the art. The above description is merely an example and is not intended to be limiting in any way. The present invention is limited only as defined in the following claims and the equivalents thereto.

DESCRIPTION OF SYMBOLS 1 life pattern prediction apparatus 101 position information storage part 102 teacher data group 11 action type extraction part 12 life pattern type assignment part 13 life pattern type prediction part 14 machine learning engine 15 life pattern type verification part 16 action time prediction part 17 action time verification Part 2 Mobile terminal

Claims (14)

In the life pattern prediction apparatus for predicting the life pattern of the future period of Yu over The,
A machine learning engine constructed using as teaching data the time series life pattern types for each unit period collected from an unspecified number of people;
Position information storage means for storing position information of the portable terminal and time series data of day and time;
Action type extraction means for extracting an action type based on stay from the time series data ;
Life pattern type assignment means for assigning any given life pattern type based on one or more action types for each unit period;
A life style characterized by having life pattern type prediction means for inputting a time series life pattern type in a past period (> unit period) to the machine learning engine and predicting a life pattern type of a future period as its output. Pattern prediction device. The life pattern predicting apparatus according to claim 1, wherein the life pattern type assigning means assigns the life pattern type based on a combination of a plurality of action types in the unit period. The life pattern predicting apparatus according to claim 1 or 2, wherein the life pattern type assigning means assigns the life pattern type based on a transition of a plurality of action types in the unit period. The life pattern type prediction means predicts the life pattern type of the future period of the user based on the appearance probability of each life pattern type in the past period. The life pattern prediction device described in. The life pattern prediction device according to claim 4, wherein the life pattern type prediction means calculates an appearance probability for each life pattern type based on periodicity of a plurality of life pattern types in the past period. . After the start of the future period, the life of verifying whether the action type extracted by the action type extraction unit matches the action type constituting the life pattern type predicted by the life pattern type prediction unit The life pattern prediction apparatus according to any one of claims 1 to 5, further comprising pattern type verification means. The action type extraction means further extracts start and end times of each action type;
Based on the life pattern type predicted by the life pattern type prediction means, the action type extracted by the action type extraction means, and the start / end time of the start / end time of the action type start / end in the future period of the user The life pattern prediction device according to any one of claims 1 to 6, further comprising action time prediction means for predicting time. 8. The action time prediction means according to claim 7, wherein the distribution of the start / end time of the action type in the life pattern type calculated to have the highest appearance probability by the life pattern type prediction means is applied. Life pattern prediction device. The action time prediction means is
The distribution of the start / end time of the action type is calculated for each life pattern type,
9. The life pattern prediction device according to claim 7, wherein prediction is performed based on the appearance probability for each life pattern type and the probability due to the distribution of the start / end times of the action type. An action for verifying whether the start / end time of the action type extracted by the action type extraction means matches the start / end time of the action type predicted by the action time prediction means after the start of the future period The life pattern prediction device according to any one of claims 7 to 9, further comprising time verification means. The unit period is a day,
The past period is a continuous day,
The future period is the next day,
In the teacher data of the machine learning engine, the day of the week is associated with the daily time-series life pattern type,
The life pattern type prediction means, to the machine learning engine, claims 1 to enter the life pattern type of time-series in the past consecutive days, and wherein the predicting the life pattern type of the next day as an output The life pattern prediction device according to any one of 10 . In the mobile terminal to predict the life pattern of the future period of Yu over The,
A machine learning engine constructed using as teaching data the time series life pattern types for each unit period collected from an unspecified number of people;
Position information storage means for storing position information of the portable terminal and time series data of day and time;
Action type extraction means for extracting an action type based on stay from the time series data ;
Life pattern type assignment means for assigning any given life pattern type based on one or more action types for each unit period;
A portable phone characterized by further comprising life pattern type prediction means for inputting into the machine learning engine the time series life pattern types in the past period (> unit period) and predicting life pattern types in the future period as its output. Terminal. In a program that causes a computer mounted on a device to predict a life pattern of a user's future period,
A machine learning engine constructed using as teaching data the time series life pattern types for each unit period collected from an unspecified number of people;
Position information storage means for storing position information of the portable terminal and time series data of day and time;
Action type extraction means for extracting an action type based on stay from the time series data ;
Life pattern type assignment means for assigning any given life pattern type based on one or more action types for each unit period;
The feature is that the computer functions as life pattern type prediction means for inputting time series life pattern types in the past period (> unit period) into the machine learning engine and predicting life pattern types in the future period as its output. The program to In a device life pattern prediction method for predicting a life pattern of a user's future period,
The device
A machine learning engine constructed using as teaching data the time series life pattern types for each unit period collected from an unspecified number of people;
A position information storage unit that time accumulated series data of the position information and the date time of the portable terminal has a <br/>,
The apparatus comprising
A first step of extracting an activity type based on stay from the time-series data;
A second step of assigning any predetermined life pattern type based on one or more action types for each unit period;
An apparatus comprising : a third step of inputting time-series life pattern types in a past period (> unit period) into the machine learning engine and predicting life pattern types of a future period as its output . Life pattern prediction method.

Images