Method and system for predicting the likelihood of event occurrence using time regularity function and recency-frequency function

Abstract

The present disclosure relates to a method and system of predicting the likelihood of event occurrence using a time regularity function and a recency-frequency function, and more particularly, a method and system that enables more accurate predictions than before by reflecting a residual effect depending on an event occurrence time when trying to predict the likelihood of event occurrence in the future.

Description

METHOD AND SYSTEM FOR PREDICTING THE LIKELIHOOD OF EVENT OCCURRENCE USING TIME REGULARITY FUNCTION AND RECENCY-FREQUENCY FUNCTION

[0001] The present disclosure relates to a method and system of predicting the likelihood of event occurrence using a time regularity function and a recency-frequency function, and more particularly, a method and system that enables more accurate predictions than before by reflecting a residual effect depending on an event occurrence time when trying to predict the likelihood of event occurrence in the future.

[0002] For events that occur repeatedly over a predetermined period of time, it may be possible to predict the likelihood of event occurrence during a future period, especially a short-term future period, and various methodologies are used in various fields to predict such the likelihood of event occurrence.

[0003] Meanwhile, when predicting the likelihood of event occurrence, the periodicity of the event is generally used as the most important information. In the prior art, clumpiness has been used as a key indicator when inferring periodicity, wherein the clumpiness is based only on a time difference between an occurrence frequency of an event and an occurrence time so that it can be used as a valid indicator when the effect of each event is independent, but if an event occurrence affects a subject (target), that is, if there is a residual effect from the event occurrence, then the likelihood of event occurrence in the future may change according to the residual effect, and thus the usefulness of the clumpiness in inferring periodicity may be greatly reduced.

[0004] On the other hand, a recency, frequency, monetary value (RFM) methodology is also widely used to predict the likelihood of event occurrence. Among these, the recency is an indicator that reflects information on an occurrence time of the most recently occurred event from the present time, but the indicator does not reflect any information related to events other than the most recent event, and thus there has been criticism about the limitations of the indicator.

[0005] The present disclosure is proposed in consideration of the foregoing problems, which aims to increase the accuracy of predicting the likelihood of event occurrence by generating a periodicity and time regularity function that reflects a residual effect in predicting event occurrence and indexes the results accordingly, and generating a recency-frequency function that reflects the residual effect and indexes the results in order to overcome the limitation of the recency indicator in the existing RFM methodology.

[0006] An aspect of the present disclosure is to predict a likelihood that an arbitrary event will occur in the future, but to use a functional equation that can take into account the event regularity or event periodicity in a measurement period based on a residual effect from the occurrence of the event, thereby allowing more accurate prediction than before.

[0007] In addition, an aspect of the present disclosure is to increase the prediction accuracy of event occurrence by using a functional expression that takes into account the recency and frequency of event occurrence in the measurement period based on the residual effect.

[0008] Technical problems of the present disclosure are not limited to the above-mentioned problems, and other technical problems which are not mentioned herein will be clearly understood by those skilled in the art from the description below.

[0009] The present disclosure is contrived to solve the foregoing problems, and a method of predicting the likelihood of event occurrence of a specific event according to the present disclosure may include acquiring history information in which the event has occurred a plurality of times during a certain period; calculating a residual effect value according to each occurrence time of the event; calculating a regularity ratio based on the residual effect value; and calculating the likelihood of event occurrence of the specific event based on the regularity ratio.

[0010] Furthermore, in the method of predicting the likelihood of event occurrence, the residual effect value may be calculated by an equation defined as the sum of a sequence including at least two or more terms, wherein each term of the sequence includes an iterator that increases by a predetermined increment in the denominator.

[0011] Furthermore, in the method of predicting the likelihood of event occurrence, the residual effect value may be calculated using [Equation 1].

[0012] [Equation 1]

[0013]

[0014] (L: residual effect value; T: event occurrence observation duration; t: event occurrence time; i: iterator from 1 to wT; w is real number; i is natural number)

[0015] Furthermore, in the method of predicting the likelihood of event occurrence, the regularity ratio may be calculated using [Equation 2].

[0016] [Equation 2]

[0017]

[0018] (TR: regularity ratio; R: sum of maximum residual effects occurring at each operation iterator time from i = 1 to i = wT; Rmax: R value calculated when an event occurrence occurs at all times from i = 1 to i = T; w is a real number; i is a natural number)

[0019] Meanwhile, a method of predicting the likelihood of event occurrence of a specific event according to another embodiment of the present disclosure may include acquiring history information in which the event has occurred a plurality of times during a certain period; calculating a residual effect value according to each occurrence time of the event; calculating a recency-frequency value according to the residual effect value of the event; and calculating the likelihood of event occurrence of the specific event based on the recency-frequency value.

[0020] Furthermore, in the method of predicting the likelihood of event occurrence, the recency-frequency value may be calculated by an equation defined as the sum of a sequence including at least two or more terms, wherein each of the terms includes an iterator that increases by a predetermined increment in the denominator.

[0021] Furthermore, in the method of predicting the likelihood of event occurrence, the recency-frequency value may be calculated using [Equation 3].

[0022] [Equation 3]

[0023]

[0024] (RF: recency-frequency value; n: number of events; j: iterator from 1 to n; T: event occurrence observation duration; i: iterator from 1 to wT; w: weight for calculation period; tj: occurrence time of event j; Pj: event influence at tj)

[0025] Meanwhile, in a computing device for predicting the likelihood of event occurrence according to still another embodiment of the present disclosure, the computing device may include a central processing unit and a memory, and the central processing unit may execute instructions for executing a method of predicting the likelihood of event occurrence stored in the memory, wherein the method of predicting the likelihood of event occurrence includes acquiring history information in which the event has occurred a plurality of times during a certain period; calculating a residual effect value according to each occurrence time of the event; calculating a regularity ratio based on the residual effect value; and calculating the likelihood of event occurrence of the specific event based on the calculated event occurrence period from the history information and the regularity ratio.

[0026] Furthermore, in the computing device, the residual effect value may be calculated by an equation defined as the sum of a sequence including at least two or more terms, wherein each of the terms includes an iterator that increases by a predetermined increment in the denominator.

[0027] Meanwhile, in a computing device for predicting the likelihood of event occurrence according to yet still another embodiment of the present disclosure, the computing device may include a central processing unit and a memory, and the central processing unit may execute instructions for executing a method of predicting the likelihood of event occurrence stored in the memory, wherein the method of predicting the likelihood of event occurrence includes acquiring history information in which the event has occurred a plurality of times during a certain period; calculating a residual effect value according to each occurrence time of the event; calculating a recency-frequency value according to the residual effect value of the event; and calculating the likelihood of event occurrence of the specific event based on the calculated event occurrence period from the history information and the recency-frequency value.

[0028] Furthermore, in the computing device, the recency-frequency value may be calculated by an equation defined as the sum of a sequence including at least two or more terms, wherein each of the terms includes an iterator that increases by a predetermined increment in the denominator.

[0029] According to the present disclosure, when an effect of an event occurrence affects a target (subject) of a next event occurrence, the residual effect may be reflected in predicting the likelihood of event occurrence, thereby having an effect of predicting a more accurate and realistic likelihood of event occurrence.

[0030] In addition, according to the present disclosure, the limitation of a recency indicator used in an RFM methodology, which does not sufficiently reflect the effect of events other than a recently occurred event, may be resolved, thereby having an effect of allowing more accurate prediction of the likelihood of event occurrence for a future period.

[0031] On the other hand, the effects of the present disclosure may not be limited to the above-mentioned effects, and other technical effects which are not mentioned herein will be clearly understood by those skilled in the art from the description below.

[0032] FIG. 1 is a diagram for conceptually illustrating a method of predicting the likelihood of event occurrence according to the present disclosure.

[0033] FIG. 2 is a flowchart showing a method of predicting the likelihood of event occurrence according to a first embodiment of the present disclosure.

[0034] FIG. 3 is a view for explaining the characteristics of an equation used to calculate a residual effect value.

[0035] FIG. 4 is a view for understanding a time regularity function with reference to a graph.

[0036] FIG. 5 is a flowchart showing a method of predicting the likelihood of event occurrence according to a second embodiment of the present disclosure.

[0037] FIG. 6 is a view for understanding a recency-frequency function with reference to a graph.

[0038] The details of the objects and technical configurations of the present disclosure and operational effects thereof will be more clearly understood from the following detailed description based on the accompanying drawings appended hereto. Hereinafter, embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings.

[0039] Embodiments disclosed herein should not be interpreted as limiting or used to limit the scope of the present disclosure. It is apparent for those skilled in the art that a description including embodiments herein has various applications. Therefore, any embodiments described in the detailed description of the present disclosure are illustrative for better understanding of the present disclosure and are not intended to limit the scope of the present disclosure to the embodiments.

[0040] Functional blocks illustrated in the drawings and described hereunder are only examples of possible implementations. In other implementations, other functional blocks may be used without departing from the concept and scope of the detailed description. Furthermore, one or more functional blocks of the present disclosure are illustrated as separate blocks, but one or more of the functional blocks of the present disclosure may be a combination of various hardware and software elements that execute the same function.

[0041] In addition, an expression that some elements are "included" is an expression of an "open type", and the expression simply denotes that the corresponding elements are present, but should not be construed as excluding additional elements.

[0042] Moreover, in case where it is mentioned that one element is "connected" or "coupled" to the other element, it should be understood that one element may be directly connected to the other element, but another element may be present therebetween.

[0043]

[0044] FIG. 1 is a diagram for conceptually understanding a method and system of predicting the likelihood of event occurrence according to the present disclosure.

[0045] As can be seen in the drawing, the present disclosure relates to a method of predicting the likelihood of an event occurring in the future based on past information on the occurrence of a series of events, on the assumption that the calculation is executed by the computing device 100.

[0046] The present disclosure is expected to be particularly useful when analyzing an event with a residual effect, since while most of conventional event probability prediction methodologies have performed calculations for prediction based on the forced assumption that the effects of the event are independent, in the present disclosure, when an event occurs, the residual effects of the event are further taken into consideration, thereby more accurately predicting the likelihood of event occurrence in the future.

[0047] In addition, while conventional methodologies can produce meaningful results only when a target of measurement (a subject of event occurrence) is in the same environment, especially when a period of time subject to analysis is the same, the present disclosure is differentiated in that meaningful results can be produced even when the target of measurement is in a different environment, especially when the period of time subject to analysis is different. For example, assuming an environment where clinical trials for a specific drug are conducted, in some cases, drug administration times for clinical patients may not match, but even in this situation, the method of predicting the likelihood of event occurrence according to the present disclosure may be effectively applicable. As another example, assuming that a likelihood of performing, by SNS service users, a specific action (e.g. participating in or withdrawing from a specific promotion, etc.) is analyzed, while conventional methodologies can obtain meaningful results only when performing analysis on users who have the same or similar subscription time, and thus who have used the SNS service for a similar period of time, the prediction method according to the present disclosure is differentiated in that meaningful results can be obtained regardless of differences in subscription time, SNS service use period, and the like.

[0048] For reference, the method and system of predicting the likelihood of event occurrence according to the present disclosure may be applicable regardless of field as long as it calculates a likelihood, that is, a probability, of the same event occurring in any future period based on learning during a specific period, and for example, may be used in calculations to predict customer activity (customer lifetime value prediction, customer future grade prediction, customer churn likelihood prediction, customer loyalty likelihood prediction, etc.) in the retail business field, in calculations to predict a likelihood of improving the effectiveness of clinical patients during clinical trials for new drugs in the bio business field, and in calculations to predict a likelihood of viewing content at a specific point in time for each customer, and to predict a likelihood of a subscriber withdrawing from a service in the entertainment field. It is understood that fields not mentioned in this detailed description are not excluded from applicable fields as long as the technical concept described through this specification are applied.

[0049] The method for predicting the likelihood of event occurrence according to the present disclosure mentions that each step can be performed on the computing device 100, and in this case, it is assumed that the processing device 100 is equipped with a central processing unit and a memory. The central processing unit may also be referred to as a controller, a microcontroller, a microprocessor, a microcomputer, or the like. Furthermore, the central processing unit may be implemented by hardware or firmware, software, or a combination thereof, and configured to include an application specific integrated circuit (ASIC) or a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), or a field programmable gate array (FPGA) when implemented using hardware, and configured with firmware or software to include a module, a procedure, a function or the like that performs the foregoing functions or operations when implemented using firmware or software. In addition, the memory may be implemented as Read Only Memory (ROM), Random Access Memory (RAM), Erasable Programmable Read Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), flash memory, Static RAM (SRAM), a hard disk drive (HDD), a solid state drive (SSD) or the like.

[0050] In some cases, the computing device 100 may be implemented as a server, and in this case, the server may be a device that stores and executes a program, that is, a set of instructions, for actually implementing a method of predicting the likelihood of event occurrence according to the present disclosure. The type of the server may be at least one server PC managed by a specific user or may be a type of cloud server provided by another company, that is, a type of cloud server that the user can sign up to use. In addition, the computing device may be implemented in the form of a cloud system in which a server does not exist, that is, a serverless cloud system in which a plurality of distributed and individually managed computing devices operate in conjunction with one another as a single system.

[0051] Additionally, in some cases, the method of predicting the likelihood of event occurrence according to the present disclosure may be executed not on a single computing device 100 but on a computing system consisting of a plurality of computing devices. Within the computing system, a plurality of computing devices required to execute the above prediction method may be respectively set to perform different operations.

[0052] As such, the method for predicting the likelihood of event occurrence according to the present disclosure may be executed by a computing device or computing devices equipped with a central processing unit and a memory.

[0053]

[0054] The method for predicting the likelihood of event occurrence to be described in this specification will be divided into two types for convenience, which will be referred to as a first embodiment and the second embodiment, respectively. Among these, the first embodiment relates to a prediction method based on a [time regularity function], and the second embodiment relates to a prediction method based on a [recency-frequency function].

[0055] First, FIG. 2 is a flowchart showing a method of predicting the likelihood of event occurrence according to a first embodiment. Referring to the drawing, the first embodiment may include a step S101 of acquiring, by the computing device 100, event occurrence history information, a step S103 of calculating the residual effect values of the event, a step S105 of calculating a time regularity ratio of an event with periodicity, and a step S107 of predicting the likelihood of event occurrence.

[0056] The first step (S101) of acquiring event occurrence history information may be understood as a step of collecting the minimum information necessary for the computing device 100 to predict a likelihood of occurrence of an event according to the first embodiment. Any case in which the computing device 100 can load so-called history information including information on a plurality of event occurrences through the user's operation may be included in this step, and for example, downloading data including history information from the outside through a network by the user's operation, copying data including history information from a USB storage device by the user's operation, and the like may all be included as an example of step S101.

[0057] The history information may include at least information on times when events occurred and information on a period during which events occurred, and additionally, may further include information on a subject (SNS service subscriber, clinical patient, etc.) who generated the event, a description of the event (participation in a specific promotion, observation of improvement in a specific clinical outcome indicator, etc.), or information that quantifies the effect of the event occurrence (an expected profit from participation in a promotion, a degree of improvement compared to a previous clinical trial, etc.). From the above-described history information, the computing device 100 may extract a number of occurrences of events, an occurrence time, a time interval between any two events, and the like.

[0058] Subsequent to acquiring the history information, the computing device 100 may calculate a residual effect value of the event (S103). In general, when a random phenomenon according to the laws of nature appears, the effects of that phenomenon may affect the surroundings, and it is to be understood that the residual effects of the event mentioned in this detailed explanation may further include effects that can be interpreted in social scientific terms, including effects that affect the psychology of a person who is a subject of event occurrence in addition to natural scientific effects caused by the occurrence of such natural phenomena. The present disclosure is proposed to numerically reflect a residual effect that can be interpreted in this broad sense in predicting the likelihood of event occurrence in the future.

[0059] The computing device 100 may use a certain equation to calculate the residual effect value, and in this case, the equation may be defined as the sum of a sequence including at least two or more terms, and in particular in this case, each term may include an iterator that increases by a predetermined increment in the denominator. That is, by using a denominator where the iterator increases consistently, the residual effect depending on when the event occurred may be reflected in the equation.

[0060] An example of an equation for calculating the residual effect value may be represented as shown in [Equation 1] below.

[0061] [Equation 1]

[0062]

[0063] In the above [Equation 1], L is a residual effect value of the event, T is an event occurrence observation duration, t is an event occurrence time, and i is a variable that increases by a predetermined increment as an operation iterator from 1 to wT. w, which is a weight that sets the calculation period, is a real number. As can be seen in [Equation 1], in the equation for calculating a residual effect value, a residual effect according to an event occurrence time may be derived by using the iterator (i) as a denominator that increases by the increment. The residual effect may have a high value when the iterator i is at the same time as the event occurrence time, and when there is an increase in the iterator i based on a fixed specific event time, a residual effect decay slope of the event may appear because the iterator i is included in the denominator.

[0064] FIG. 3 is a diagram for more easily understanding [Equation 1] described above, wherein it is shown in FIG. 3 how the residual effect values of events with different occurrence times can be calculated differently. FIG. 3 assumes a situation when an event occurrence observation duration T is 100, wherein a total calculation period subject to analysis is set to 2T. (In this case, w = 2 in the above [Equation 1]) The residual effects according to the occurrence times of the events, which are 1, 50, and 100, are visualized, and referring to the drawing, it can be seen that when the event occurs at t = 1, 50, and 100, the residual effect values at those time points have the highest value (1.0), and then the residual effect values until the next event occurs decrease along a decay slope.

[0065] The computing device 100 may calculate a time regularity ratio based on the residual effect value calculation logic. The time regularity ratio may be understood as an indicator of how irregular the occurrence of observed events is compared to when the events occur regularly at every time during a random measurement period, and the regularity ratio may preferably be calculated through [Equation 2] below.

[0066] [Equation 2]

[0067]

[0068] In the above [Equation 2], regularity R is a sum of maximum residual effects that occurred at each calculation time from i = 1 to i = wT, which implies the distribution information of event occurrence times tiwithin an event occurrence observation duration T. Here, w, which is a weight of the calculation period, may be a real number. Rmaxis a maximum regularity R value calculated when an event occurs at any time that may occur from i = 1 to i = T, and has a maximum event distribution. Here, i indicates a specific time and may be a natural number greater than or equal to 1. A final calculated regularity ratio TR is a value obtained by dividing the regularity R, which is the sum of the maximum residual effects that occurred at each calculation time point, by Rmax.

[0069] A so-called time regularity function, which can be defined through steps S103 and S105, may calculate a maximum residual effect sum R according to an event occurrence time differently depending on a participation period of an event subject, and in other words, may allow a degree of irregularity of an event to be measured regardless of when the event subject participated in a (clinical) trial or when the event subject signed up for an SNS service.

[0070] Meanwhile, subsequent to calculating a regularity ratio, the computing device 100 may calculate the likelihood of event occurrence of a specific event using the regularity ratio (S107). The likelihood of event occurrence may be calculated as a probability value, but may not be necessarily limited thereto, and as long as it is a value whose extent can be known by comparing it to the probability value (1) that a random event can occur unconditionally, there is no limit to its form. In addition, in calculating the likelihood of event occurrence, the computing device 100 may obtain a period value from previously acquired history information, that is, history information that includes at least information on times when events occurred and information on a period during which events occurred, and then calculate the likelihood of event occurrence using an equation combining the above period value and the time regularity ratio.

[0071] FIG. 4 is a diagram for more easily understanding [Equation 2] described above, wherein it is shown in FIG. 4 how the regularity ratio can be calculated differently when the occurrence times of a plurality of events vary. FIG. 4 assumes a situation when T = 100, and a calculation period subject to analysis is set to 2T. In the drawing, an upper left graph shows a time regularity value R when events occur at time points 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, an upper center graph when events occur at time points 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, an upper right graph when events occur at time points 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, a lower left graph when events occur at time points 20, 40, 60, 80, 100, a lower center graph when events occur at time points 10, 20, 30, 40, 50, and a lower right graph when events occur at time points 60, 70, 80, 90, 100, respectively. Referring to the drawing, an area with a light gray shade is a graph showing a situation when the occurrence of an event occurs at every time point from 1 to 100, that is, when the R value is maximum, and an area with a dark gray shade is a graph showing a situation when an event corresponding to each time point occurs in [Equation 2]. The time regularity ratio is a value equal to an area of a region under a blue line compared to that under a yellow line. A time regularity ratio (TR) value calculated in each case is illustrated at the top of each graph.

[0072] In the above, the first embodiment has been described above with reference to FIGS. 2, 3, and 4.

[0073] Meanwhile, in the first embodiment, [Equation 1] and [Equation 2] have been mentioned as preferred equations, but other types of equations may also be used as long as they satisfy the definition as equations for calculating the residual effect value described above. For example, in the above [Equation 1], a square operation is included for a normalization curve, but in addition to the square operation, various methods of calculating a total amount may replace the square operation. In addition, in the above [Equation 1], the operation iterator i increases from 1 to 2T, but it can be set to various periods using an arbitrary w value, and conversely, when designing an equation using a reverse calculation method that decreases from T to 1, instead of the residual effect value, an expected effect value (a value obtained by calculating an influence when approaching an occurrence time of an event according to the event occurrence time) may be calculated.

[0074]

[0075] FIG. 5 is a flowchart showing a method of predicting the likelihood of event occurrence according to a second embodiment. Referring to the drawing, the second embodiment may include a step S201 of acquiring history information in which a specific event has occurred a plurality of times during a certain period, a step S203 of calculating a residual effect value of the event, a step S205 of calculating a recency-frequency value according to each occurrence time of the event, and a step S207 of calculating the likelihood of event occurrence of the specific event based on the recency-frequency value.

[0076] The step S201 of acquiring, by the computing device 100, history information is substantially the same as step S101 described above in the first embodiment, and thus redundant description thereof will be omitted herein.

[0077] Subsequent to step S201, the step S203 of calculating a residual effect value of the event may be executed, but this step is also substantially the same as step S103 described above, and thus redundant description thereof will be omitted herein.

[0078] Meanwhile, subsequent to step S203, the step S205 of calculating a recency-frequency value may be executed. An RFM methodology has been widely used in predicting the likelihood of event occurrence in the related art, but it is mentioned at the beginning that a recency factor among them has a limitation in that it only reflects information on the most recent event occurrence time. The second embodiment may reflect an influence of the occurrence times of the remaining events other than the most recent event occurrence time using an equation including the residual effect value calculation described in the first embodiment.

[0079] In the second embodiment, the computing device 100 may use a certain equation to calculate the recency-frequency value, and in this case, the equation may be defined as the sum of a sequence including at least two or more terms, and in particular in this case, each term may include an iterator that increases by a predetermined increment in the denominator. That is, by using a denominator where the iterator increases consistently, the residual effect depending on when the event occurred may be reflected in the equation.

[0080] An example of an equation for calculating the recency-frequency value may be represented as shown in [Equation 3] below.

[0081] [Equation 3]

[0082]

[0083] In the above [Equation 3], RF represents a recency-frequency value, n is a number of events, j is an iterator from 1 to n, T is an event occurrence observation duration, i is an iterator from 1 to wT, w is a calculation period weight, tjis an occurrence time of event j, and Pjdenotes an event influence at tj. That is, in this step, a residual effect value for each event is calculated using [Equation 1] described above (step S203), and a recency-frequency value is calculated by obtaining a total amount of residual effect values of events within an observation duration.

[0084] [Equation 3] is used to calculate a residual effect for each event according to a time period between events for a number of n events that occurred for each target during a certain period (i = 1 to wT), similar to a regularity function. The calculated residual effect value reflects recency such that the later the event occurs, the higher the value (area) is. A total amount of values calculated by summing up for n frequencies, or a value obtained by dividing and normalizing the total amount obtained in this manner by a number of n events may be defined as a new indicator reflecting recency and frequency. In other words, the above [Equation 3] does not only reflect the effect of the most recent event, but also reflects all influences (residual effects) according to multiple event occurrences, and also includes an operation capable of reflecting frequency, and therefore may be said to be an equation for calculating a new indicator that reflects both recency and frequency values. Meanwhile, the Pjis an arbitrary quantification of the influence of event j, and plays a role in calculating RFM as a single numerical value in terms of customer value analysis.

[0085] For reference, it is to be understood that the above [Equation 3] may be used in various methods or fields of calculating and summing residual effect values (areas) for multiple events, and is not limited to the methods and fields mentioned in this detailed description.

[0086] FIG. 6 is a diagram for more easily understanding [Equation 3], wherein it is shown in FIG. 6 how the residual effect value and the recency-frequency value can be calculated differently when the occurrence times of a plurality of events vary. FIG. 6 assumes a situation when T = 100, and a calculation period subject to analysis is set to 2T. In the drawing, an upper left graph shows a recency-frequency value, which is the sum of residual effect values, when events occur at time points 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, an upper center graph when events occur at time points 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, an upper right graph when events occur at time points 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, a lower left graph when events occur at time points 20, 40, 60, 80, 100, a lower center graph when events occur at time points 10, 20, 30, 40, 50, and a lower right graph when events occur at time points 60, 70, 80, 90, 100, respectively. Referring to the drawing, if all the occurrence areas under a line are summed, then a recency-frequency (RF) value calculated by [Equation 3] may be obtained, wherein the recency-frequency (RF) value in each graph is illustrated at the top of the graph.

[0087] In the above, a method of predicting the likelihood of event occurrence according to the present disclosure and a system therefor have been described. Meanwhile, the present disclosure is not limited to the foregoing specific embodiments and application examples, it will be of course understood by those skilled in the art that various modifications may be made without departing from the gist of the present disclosure as defined in the following claims, and it is to be noted that those modifications should not be understood individually from the technical concept and prospect of the present disclosure.

Claims

1.A method of predicting a likelihood of event occurrence of a specific event, the method comprising:acquiring history information in which the event has occurred a plurality of times during a certain period;calculating a residual effect value according to each occurrence time of the event;calculating a regularity ratio based on the residual effect value; andcalculating the likelihood of event occurrence of the specific event based on the regularity ratio.2.The method of claim 1, wherein the residual effect value is calculated by an equation defined as the sum of a sequence including at least two or more terms, andwherein each term of the sequence includes an iterator that increases by a predetermined increment in the denominator.3.The method of claim 2, wherein the residual effect value is calculated using [Equation 1].[Equation 1](L: residual effect value; T: event occurrence observation duration; t: event occurrence time; i: iterator from 1 to wT; w is real number; i is natural number)4.The method of claim 3, wherein the regularity ratio is calculated using [Equation 2].[Equation 2](TR: regularity ratio; R: sum of maximum residual effects occurring at each operation iterator time from i = 1 to i = wT; Rmax: R value calculated when an event occurrence occurs at all times from i = 1 to i = T; w is a real number; i is a natural number)5.A method of predicting a likelihood of event occurrence of a specific event, the method comprising:acquiring history information in which the event has occurred a plurality of times during a certain period;calculating a residual effect value according to each occurrence time of the event;calculating a recency-frequency value according to the residual effect value of the event; andcalculating the likelihood of event occurrence of the specific event based on the recency-frequency value.6.The method of claim 5, wherein the recency-frequency value is calculated by an equation defined as the sum of a sequence including at least two or more terms, andwherein each of the terms includes an iterator that increases by a predetermined increment in the denominator.7.The method of claim 6, wherein the recency-frequency value is calculated using [Equation 3].[Equation 3](RF: recency-frequency value; n: number of events; j: iterator from 1 to n; T: event occurrence observation duration; i: iterator from 1 to wT; w: weight for calculation period; tj: occurrence time of event j; Pj: event influence at tj)8.A computing device for predicting a likelihood of event occurrence, the computing device comprising a central processing unit and a memory,wherein the central processing unit executes instructions for executing a method of predicting the likelihood of event occurrence stored in the memory, andwherein the method of predicting the likelihood of event occurrence comprises:acquiring history information in which the event has occurred a plurality of times during a certain period;calculating a residual effect value according to each occurrence time of the event;calculating a regularity ratio based on the residual effect value; andcalculating the likelihood of event occurrence of the specific event based on the calculated event occurrence period from the history information and the regularity ratio.9.The computing device of claim 8, wherein the residual effect value is calculated by an equation defined as the sum of a sequence including at least two or more terms, andwherein each of the terms includes an iterator that increases by a predetermined increment in the denominator.10.A computing device for predicting a likelihood of event occurrence, the computing device comprising a central processing unit and a memory,wherein the central processing unit executes instructions for executing a method of predicting the likelihood of event occurrence stored in the memory, andwherein the method of predicting the likelihood of event occurrence comprises:acquiring history information in which the event has occurred a plurality of times during a certain period;calculating a residual effect value according to each occurrence time of the event;calculating a recency-frequency value according to the residual effect value of the event; andcalculating the likelihood of event occurrence of the specific event based on the calculated event occurrence period from the history information and the recency-frequency value.11.The computing device of claim 10, wherein the recency-frequency value is calculated by an equation defined as the sum of a sequence including at least two or more terms, andwherein each of the terms includes an iterator that increases by a predetermined increment in the denominator.

Images