Prediction device, prediction program and prediction method

Abstract

To provide a prediction device, a prediction program and a prediction method which can enhance accuracy of prediction of a problem set in a test, and can reduce a learning burden on an examinee.SOLUTION: A prediction device includes: a question data reception part for receiving question data including presence / absence of a question of a contention in each continuous time before a prediction object time, for each contention of correct answer choices of an alternative type question of a plurality of tests in the past; a prediction model creation part for creating a prediction model indicating a relation between presence / absence of a question in a plurality of tests performed before the prediction object time, and the possibility of a question in the prediction object type, by a plurality of data sets of presence / absence of a question in a plurality of tests performed before a specific time, and presence / absence of a question in the specific time; and a prediction processing part for predicting the possibility of the question of the test of the prediction object time, on the basis of presence / absence of the question in the plurality of tests including a test after the latest time in the plurality of tests performed before the specific time, and the prediction model.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a prediction device, a prediction program, and a prediction method. [Background technology]

[0002] National exams and other qualification exams are usually held periodically, from once to several times a year. Preparatory schools and other institutions that provide preparation for these exams work to predict the questions that will appear on the exams. The results of these predictions are used for exam preparation and the creation of mock exams. Preparatory schools and other institutions that make highly accurate predictions have a good reputation, and the accuracy of their predictions can affect the number of students they receive.

[0003] When predicting exam questions, prep school instructors classify past exam questions by subject area, extract past question trends, such as the frequency and frequency of questions in each subject area, from the classification results, and predict the subjects and questions that will appear in the next exam based on the past question trends. However, the number of past questions can be enormous, which requires a significant workload. Furthermore, exam question predictions depend to some extent on the experience, ability, and intuition of the prep school instructor, making prediction accuracy unstable.

[0004] Therefore, for example, Patent Document 1 proposes a prediction device in which a test question analysis unit classifies questions into multiple question fields based on the question text and answer options, and uses information on question trends indicating the number of questions for each question field for each test as question data, a prediction method selection unit selects an applicable prediction model format based on the question trends indicating the number of questions for each question field, and this prediction processing unit predicts the question trends indicating the number of questions for each question field in future tests.The prediction device in Patent Document 1 can predict the trend in the number of questions for each field, so the prediction results can be used to determine which field should be focused on for study or to determine the number of questions for each field when creating mock tests. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2020-9129 Summary of the Invention [Problem to be solved by the invention]

[0006] As such, the prediction device disclosed in Patent Document 1 can predict question trends for each field, but cannot predict the specific types of questions that will be asked. Therefore, with only the prediction device disclosed in Patent Document 1, examinees would have to comprehensively study at least the fields that are judged to be important, and the effect of reducing the study burden on examinees is not significant.

[0007] The present invention has been made in consideration of the above-mentioned circumstances, and aims to provide a prediction device, a prediction program, and a prediction method that can improve the accuracy of predicting questions that will appear on an exam and reduce the learning burden on examinees. [Means for solving the problem]

[0008] The present inventor has proposed a prediction device that predicts the likelihood of multiple-choice questions being asked in a future target test based on question trends in past tests, the prediction device comprising: a question data receiving unit that receives question data for each point of a correct answer option that is the correct answer to a multiple-choice question in multiple past tests, including whether the point was asked in each of consecutive tests prior to the target test; and a question data receiving unit that receives question data for each point of a correct answer option, including whether the point was asked in multiple tests prior to the target test, based on multiple data sets including whether the point was asked in multiple tests prior to the target test and whether the point was asked in the specific test. The present inventors have found that a prediction device equipped with a prediction model creation unit that creates a prediction model indicating the relationship between the likelihood of a question being asked in a target exam and / or an index related to the likelihood of the question being asked, and a prediction processing unit that, for each correct answer question, predicts the likelihood of the question being asked in a target exam or calculates an index related to the likelihood of the question being asked, based on the prediction model and whether the question has been asked in multiple exams, including exams after the latest of multiple exams held before a specific exam, contained in the question data.

[0009] (1) A prediction device that predicts the likelihood of multiple-choice questions being asked in a future target test based on past test question trends, a question data receiving unit that receives question data including whether or not a question on each of the points of the correct answer options that are correct answers to the multiple-choice questions in the past multiple exams has been asked in each of the past consecutive exams prior to the exam to be predicted; a prediction model creation unit that creates, for each point of the correct answer, a prediction model that indicates the relationship between the presence or absence of a question in multiple exams taken before the target time to be predicted and the possibility of the question being asked in the target time to be predicted and / or an index related to the possibility of the question being asked, from multiple data sets of whether the question was asked in multiple exams taken before the specific time and whether the question was asked in the specific time, which are included in the question data; a prediction processing unit that predicts the possibility of the question being asked in the exam to be predicted, or calculates an index related to the possibility of the question being asked, based on whether or not the question has been asked in multiple exams, including the exam after the latest of the multiple exams held before the specific exam, which are included in the question data, and the prediction model, for each point of the correct answer; A prediction device comprising:

[0010] (2) The prediction model is a logistic regression model in which the presence or absence of the question in multiple exams conducted before the prediction target exam is used as an explanatory variable, The prediction device according to (1), wherein the prediction processing unit calculates the probability that the point of the correct answer will be asked.

[0011] (3) The prediction model is a logistic regression model in which the presence or absence of the question in multiple exams conducted before the prediction target exam is used as an explanatory variable, The prediction device according to (1), wherein the prediction processing unit determines the learning priority of the point of the correct answer.

[0012] (4) A prediction program that predicts the likelihood of multiple-choice questions being asked in a future target test based on past test question trends, The computer is caused to function as a question data receiving unit, a prediction model creating unit, and a prediction processing unit; the question data receiving unit receives, for each point of a correct answer that is a correct answer to the multiple-choice question in the past multiple exams, question data including whether or not the point has been asked in each of the past consecutive exams prior to the prediction target exam; The prediction model creation unit creates a prediction model showing the relationship between the presence or absence of a question in a plurality of exams conducted before the prediction target exam and the possibility of the question being asked in the prediction target exam and / or an index related to the possibility of the question being asked, based on a plurality of data sets of whether the question was asked in a plurality of exams conducted before a specific exam and whether the question was asked in the specific exam, which are included in the question data, for each point of the correct answer; The prediction processing unit predicts the possibility of the question being asked in the exam to be predicted, or calculates an index related to the possibility of the question being asked, based on whether the question has been asked in a plurality of exams, including an exam after the latest of the plurality of exams held before the specific exam, which are included in the question data, and the prediction model, for each point of the correct answer; Prediction program.

[0013] (5) A prediction method for predicting the likelihood of multiple-choice questions being asked in a future target test based on past test question trends, comprising: a question data receiving step for receiving question data including whether or not a question on each of the points of the correct answer options that are correct answers to the multiple-choice questions in the past multiple examinations has been asked in each of the past consecutive examinations prior to the examination to be predicted; a prediction model creation process for creating a prediction model that shows the relationship between the presence or absence of a question in a plurality of exams conducted before the target test and the likelihood of the question being asked in the target test and / or an index related to the likelihood of the question being asked, based on multiple data sets of whether the question was asked in a plurality of exams conducted before the target test and whether the question was asked in the target test, for each of the points of the correct answer; a prediction processing step of predicting the possibility of the question being asked in the exam to be predicted, or calculating an index related to the possibility of the question being asked, based on whether or not the question has been asked in a plurality of exams, including an exam after the latest of a plurality of exams held before the specific exam, which are included in the question data, and the prediction model, for each point of the correct answer; A forecasting method including:

[0014] According to the present invention, it is possible to improve the accuracy of prediction of questions that will appear on an exam and to reduce the study burden on examinees. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a schematic diagram of a prediction device according to an embodiment of the present invention. [Figure 2] FIG. 2 is a schematic diagram of an example of question data used in the prediction device according to the present embodiment. [Figure 3] FIG. 1 is a schematic diagram illustrating a hardware configuration of a prediction device according to an embodiment of the present invention. [Figure 4] FIG. 2 is a flow diagram of a prediction method according to the present embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0016] Hereinafter, embodiments of the present invention will be described, but the present invention is not limited to the description of the embodiments and can be practiced with appropriate modifications.

[0017] The program for realizing the software of this embodiment may be provided as a non-temporary recording medium that can be read by a computer, or may be provided in a state that can be downloaded from an external server, or may be provided so that such a program is started on an external computer and its functions are realized on a client terminal (so-called cloud computing).

[0018] In this embodiment, the term "unit" may include, for example, a combination of hardware resources implemented by a circuit in the broad sense and software information processing that can be specifically realized by these hardware resources. In addition, this embodiment handles various types of information, which may be represented by, for example, physical values ​​of signal values ​​representing voltage and current, high and low signal values ​​as a binary bit set consisting of 0 or 1, or quantum superposition (so-called quantum bits), and communication and calculations may be performed on a circuit in the broad sense.

[0019] Furthermore, a "circuit in the broad sense" refers to a circuit realized by at least an appropriate combination of a circuit, circuitry, processor, memory, etc. In other words, it includes an application specific integrated circuit (ASIC), a programmable logic device (e.g., a simple programmable logic device (SPLD), a complex programmable logic device (CPLD), and a field programmable gate array (FPGA)), etc.

[0020] <Prediction Device> The prediction device according to this embodiment predicts the likelihood of a multiple-choice question being asked in a future target exam based on question trends in past exams. Specifically, the prediction device includes a question data receiving unit that receives, for each point of a correct answer option that is the correct answer to a multiple-choice question in multiple past exams, question data including whether the point was asked in each consecutive exam prior to the target exam; a prediction model creating unit that creates, for each point of a correct answer option, a prediction model indicating the relationship between the question's presence in multiple exams prior to the target exam and the likelihood of the question being asked in the target exam and / or an index related to the likelihood of the question being asked in the target exam, based on the question data including whether the point was asked in multiple exams prior to the specific exam, and the prediction model; and a prediction processing unit that predicts the likelihood of the question being asked in the target exam based on the prediction model and the likelihood of the question being asked in multiple exams prior to the specific exam, including the exam after the latest of the multiple exams prior to the specific exam, based on the question data.

[0021] Fig. 1 is a schematic diagram of a prediction device according to this embodiment. The prediction device 1 shown in Fig. 1 mainly includes a question data receiving unit 11, a prediction model creating unit 12, a prediction processing unit 13, a display control unit 14, a display unit 15, and a storage unit 16.

[0022] For example, a user U of the prediction device 1, such as a cram school teacher, uses the prediction device 1 to predict questions that may appear in the next exam based on questions that appeared in past exams. Specifically, the user U extracts the correct answers for questions that appeared in past exams in advance, and calculates the question data for each topic by counting the number of times the question appeared. A prediction model is created based on this. Then, by inputting information about more recent exams into the prediction model, the prediction device 1 predicts the likelihood of multiple-choice questions appearing in future exams, or calculates an index related to the likelihood of the question appearing.

[0023] Such a prediction device 1 has high accuracy in predicting questions that will appear on the exam, allowing examinees to understand the key points they should focus on studying, further reducing the burden of studying on the examinees.

[0024] Here, "exam" refers to an exam that is repeatedly conducted for the same purpose. Examples include exams that are conducted periodically to obtain the same qualification, such as the bar exam or the tax accountant exam, and exams that are conducted periodically to enter the same school, such as high school entrance exams or university entrance exams. An exam may be conducted repeatedly multiple times, regardless of the time or frequency of the exam, whether it is held regularly or irregularly. Furthermore, an "exam" may include multiple-choice questions at least in part, and may also include other types of questions, such as essay questions or fill-in-the-blank questions, or the entire exam may be composed of multiple-choice questions.

[0025] A "correct answer for a multiple-choice question" is an answer that becomes correct when selected, regardless of whether the sentence explaining that answer is correct or incorrect. In other words, if the question requires you to select the correct answer, the "correct answer for a multiple-choice question" is the answer containing the correct sentence, and if the question requires you to select the incorrect answer, the "correct answer for a multiple-choice question" is the answer containing the incorrect sentence. In addition, for questions requiring you to select the number of correct or incorrect answers from all the answers found in exams such as the patent attorney exam or real estate agent exam, all of the correct and incorrect answers, respectively, can be considered "correct answers for the multiple-choice question." For example, if there are three correct answers in a question requiring you to select the correct number of answers, all three answers would be considered "correct answers for the multiple-choice question."

[0026] "Consecutive exams" refers to consecutive exams that have already been held in the past. For example, if the exam is held once a year, it refers to the combination of the Reiwa 2 exam, the Reiwa 3 exam, and the Reiwa 4 exam.

[0027] This section explains "an exam that is later than the latest of multiple exams held before a specific exam that are included in the question data." For example, if three datasets are used, each for an exam that is held once a year, namely, a dataset in which the specific exam is in 2022 and multiple exams are from 2020 to 2011, a dataset in which the specific exam is in 2021 and multiple exams are from 2019 to 2010, and a dataset in which the specific exam is in 2020 and multiple exams are from 2018 to 2009, the "latest of multiple exams held before a specific exam" would be the exam in 2020, and the "exam that is later than the latest of multiple exams held before a specific exam that are included in the question data" would be an exam from 2021 onwards.

[0028] The components of the prediction device 1 and their operations will be described below.

[0029] [Question Data Reception Department] The question data receiving unit 11 receives question data including, for each question point of the correct answer option that is the correct answer to a multiple-choice question in multiple past exams, whether or not the question point was asked in each of the consecutive past exams prior to the exam to be predicted.

[0030] (Question data) The question data includes, for each point of the correct answer that is the correct answer for a multiple-choice question in a number of past exams, whether or not that point has been asked in each consecutive past exam prior to the exam to be predicted. The question data received in this way is stored in a memory unit (not shown).

[0031] FIG. 2 is a schematic diagram of an example of question data used in the prediction device according to this embodiment. FIG. 2 is created for the purpose of explaining the present invention, using questions from the construction management engineer examination as a model, and differs from the actual questions in terms of the year of questioning, etc. Also, for the sake of simplicity, letters from columns A to T have been added. As shown in FIG. 2, in one embodiment, the question data may be provided in a table format.

[0032] Column A of the question data shown in Figure 2 lists the actual questions. Questions are predicted for each column, i.e., for each row of the question data. Column B lists the higher-level concept or major category of the category to which the question belongs. Column C lists the numerical value and major category ID corresponding to the major category. Column D lists the lower-level concept or minor category of the category to which the question belongs. Column E lists the numerical value and minor category ID corresponding to the minor category. Columns F to T list the number of questions for the given topic in each exam from 2002 (Reiwa 4) to 2008 (Heisei 20). Based on the question availability for each year listed in columns F to T, a dataset is created showing whether the topic was available in multiple exams held before a specific exam and whether the topic was available in a specific exam. Based on this dataset, the prediction model creation unit 12 creates a prediction model. Additionally, whether the topic was available in multiple exams held before a specific exam, including those after the most recent exam, is also extracted from this data.

[0033] The question data thus received by the question data receiving unit 11 is stored in the storage unit 16.

[0034] Using the above question data and the prediction model described below, multiple-choice questions that may be asked in the exam to be predicted are predicted.

[0035] [Prediction Model Creation Department] The prediction model creation unit 12 creates a prediction model that indicates the relationship between the presence or absence of a question in multiple exams taken before the target exam and the likelihood of the question being asked in the target exam and / or an index related to the likelihood of the question being asked, for each point of the correct answer, from multiple data sets, including the presence or absence of the question in multiple exams taken before the specific exam and the presence or absence of the question in the specific exam, which are included in the question data. The prediction model created in this way is stored in a memory unit (not shown).

[0036] (Prediction model) The prediction model shows the relationship between whether the correct answer question has been asked in multiple exams held prior to the exam being predicted, and the likelihood of the question being asked in the exam being predicted and / or indicators related to the likelihood of the question being asked.

[0037] The prediction model is not particularly limited as long as it shows the relationship between whether or not a question was asked in multiple exams held before the predicted exam and the likelihood of the question being asked in the predicted exam and / or an index related to the likelihood of the question being asked. However, in one embodiment, it is preferable to use a logistic regression model in which the explanatory variable is whether or not a question was asked in multiple exams held before the predicted exam.

[0038] Specifically, when a logistic regression model is used as a prediction model, in which the explanatory variable is whether or not a question has been asked in the past, the probability of a question being asked can be expressed by the following equations (1) and (2), where P is the probability of a question being asked.

number

[0039] First, for each point of the correct answer, a logistic regression analysis was performed on the relationship between whether or not the question was asked in multiple exams conducted before a specific exam and whether or not the question was asked in a specific exam. n Calculate.

[0040] Although not a required configuration, the classification of the correct answer's argument may be added as an explanatory variable. In that case, the following formula (3), for example, can be used instead of formula (2).

number

[0041] Equation (3) is the result of adding the third term on the right side compared to equation (2), and takes into account the contribution of the classification ID to the probability of being asked, P. A logistic regression analysis was similarly performed for each correct answer question, and β0 and β n and β' n As the category ID, for example, the major category ID shown in column C of FIG. 2 or the minor category ID shown in column D can be used.

[0042] The prediction model created in this way is stored in the storage unit 16.

[0043] [Prediction processing section] The prediction processing unit 13 predicts the possibility of a question being asked in the exam to be predicted, or calculates an index related to the possibility of a question being asked, for each point of the correct answer, based on the presence or absence of the question in multiple exams, including the exam after the latest of the multiple exams held before the specific exam, contained in the question data, and a prediction model.

[0044] In this prediction processing unit 13, for each point of the correct answer for the round to be predicted, the possibility of the question being asked is calculated using the explanatory variables of whether or not the question was asked in multiple exams held before the round to be predicted, and the prediction model described above.

[0045] When formula (1) is used, the probability P of being asked in the formula is calculated as a probability ranging from 0 to 1. The probability P of being asked as a question may be expressed as a percentage by multiplying it by 100. In this way, for each point of the correct answer option that is the correct answer for multiple-choice questions in multiple past exams, the probability of that point being asked in the question is calculated based on the prediction model.

[0046] Furthermore, for example, the value of the likelihood of being asked in a question P may be divided into several categories, and the study priority of a point belonging to a category with a higher likelihood of being asked in a question P may be determined to be higher. In this way, for each point of the correct answer option that is the correct answer to the multiple-choice question in the past multiple exams, the study priority of the point of the correct answer option may be determined based on the prediction model.

[0047] The question possibility or related indicators created in this way are stored in the storage unit 16.

[0048] [Display control unit] The display control unit 14 controls the display of the probability of a question being asked in a subsequent exam session predicted by the prediction processing unit 13, or an index related to the probability of a question being asked calculated by the prediction processing unit 13, on the display unit 15 described below.

[0049] [Display] The display unit 15 displays the possibility of a question being asked or an indicator related thereto, and may be, for example, a display.

[0050] [Hardware configuration of the prediction device] FIG. 3 is a schematic diagram showing the hardware configuration of the prediction device according to this embodiment.

[0051] 3, the prediction device 1 has a communication unit 17, a storage unit 16, and a control unit 18, and these components are electrically connected via a communication bus 19 inside the prediction device 1. These will be described in more detail below.

[0052] The memory unit 16 stores the various pieces of information described above. This is implemented as a storage device such as a solid state drive (SSD), or as a memory such as a random access memory (RAM) that stores temporarily required information (arguments, arrays, etc.) related to program calculations. The memory unit 16 may also be a combination of these. The memory unit 16 also stores various programs that can be read by the control unit 18, which will be described later.

[0053] The communication unit 17 is preferably a wired communication means such as USB, IEEE1394, Thunderbolt, or wired LAN network communication, but may also include wireless LAN network communication, mobile communication such as 3G / LTE / 5G, Bluetooth (registered trademark), etc., as needed. In other words, it is preferably implemented as a collection of multiple communication means. This allows information and commands to be exchanged between the prediction device 1 and other devices with which it can communicate.

[0054] The control unit 18 processes and controls the overall operations related to the prediction device 1. The control unit 18 is, for example, a central processing unit (CPU, not shown). The control unit 18 realizes various functions related to the prediction device 1 by reading out predetermined programs stored in the storage unit 16. That is, information processing by software (stored in the storage unit 16) is specifically realized by hardware (the control unit 18), and is executed as each functional unit in the control unit 18 shown in FIG. 1. Note that while FIG. 3 shows a single control unit 18, the actual configuration is not limited to this example, and multiple control units 18 may be provided for each function, or a single control unit may be combined with multiple control units.

[0055] <Prediction Program> The prediction program according to this embodiment predicts the likelihood of a multiple-choice question being asked in a future target exam based on question trends in past exams. Specifically, this prediction program causes a computer to function as a question data receiving unit, a prediction model creating unit, and a prediction processing unit. The question data receiving unit receives question data, including whether or not a question was asked in each consecutive exam prior to the target exam, for each correct answer option in the multiple-choice questions in multiple past exams. The prediction model creating unit creates a prediction model showing the relationship between whether or not a question was asked in multiple exams prior to the target exam and the likelihood of the question being asked in the target exam and / or indicators related to the likelihood of the question being asked in the target exam, based on multiple data sets, including the question data, for each correct answer option in the multiple exams prior to the target exam and the likelihood of the question being asked in the target exam. Furthermore, the prediction processing unit predicts the possibility of each point in the correct answer being included in the exam to be predicted, or calculates an index related to the possibility of the point being included, based on the presence or absence of the point being included in multiple exams, including the exam that was held after the latest of the multiple exams held before the specific exam, contained in the question data, and the prediction model.

[0056] The question data receiving unit, prediction model creation unit, and prediction processing unit in the prediction program may be similar to the question data receiving unit 11, prediction model creation unit 12, and prediction processing unit 13 in the prediction device 1. The prediction program may also include a display control unit, and the display control unit in the prediction program may be similar to the display control unit 14 of the prediction device 1.

[0057] <Prediction method> The prediction method of this embodiment is a prediction method that predicts the possibility of multiple-choice questions being asked in a future target prediction exam based on question trends in past exams. Specifically, this prediction method includes a question data receiving step for receiving, for each point of a correct answer option that is the correct answer to a multiple-choice question in multiple past exams, question data including whether or not the point was asked in each of the consecutive exams prior to the exam to be predicted; a prediction model creating step for creating, for each point of a correct answer option, a prediction model showing the relationship between whether or not the point was asked in multiple exams held prior to the exam to be predicted and the likelihood of the point being asked in the exam to be predicted and / or an index related to the likelihood of the point being asked in the exam to be predicted, based on the prediction model and whether or not the point was asked in multiple exams held prior to the specific exam, including the exam that was later than the latest of the multiple exams held prior to the specific exam, included in the question data; and a prediction processing step for predicting the likelihood of the point being asked in the exam to be predicted, or calculating an index related to the likelihood of the point being asked in the exam to be predicted, based on the prediction model and whether or not the point was asked in multiple exams held prior to the specific exam, included in the question data.

[0058] 4 is a flow diagram of the prediction method of this embodiment. This prediction method can be implemented by, for example, the prediction device 1 described above.

[0059] (Step S1: Question data reception process) Step S1 is a question data receiving step, which receives question data including, for each question point of the correct answer option that is the correct answer for the multiple-choice question in multiple past exams, whether or not the question point was asked in each consecutive past exam prior to the exam to be predicted. The question data only needs to include at least the past exams in which each question point was asked, and for example, the question data shown in Figure 3 can be used.

[0060] (Step S2: Prediction model creation process) Step S2 is a prediction model creation process, in which, for each point in the correct answer, a prediction model is created that shows the relationship between whether the question was asked in multiple exams held before the predicted exam and the likelihood of the question being asked in the predicted exam and / or indicators related to the likelihood of the question being asked, based on multiple data sets containing whether the question was asked in multiple exams held before the specific exam and whether the question was asked in the specific exam, which are included in the question data.

[0061] Below, we will explain how to create a specific prediction model for an exam held once a year, assuming that the target exam is the 2023 exam. The five specific exams are defined as the 2022, 2021, 2020, 2019, and 2018 exams. For multiple exams prior to the specific exam, for example, the 10 exams (10 years) prior to the specific exam are extracted (for example, if the 2022 exam is the specific exam, then 2021 to 2013, and if the 2023 exam is the specific exam, then 2020 to 2012). Five datasets are then obtained, with the specific exams being the five exams from 2022 to 2018. Using these five datasets as input, a logistic regression analysis is performed on the relationship between whether a specific exam (year) was asked (for example, 1 if the question was asked, 0 if the question was not asked) and the exams in the previous multiple exams (years) in which the question was asked. This allows us to calculate the β0 and β in Equation (1). n ,β' m can be obtained, and the prediction model expressed by equations (1) and (2) can be obtained.

[0062] In the above example, assuming that the specific test is in fiscal year 2022, the multiple tests prior to the specific test are those going back from fiscal year 2021, i.e., going back from the test immediately preceding the specific test. However, the multiple tests prior to the specific test do not necessarily have to go back from the test immediately preceding the specific test; they can go back a certain number of times, such as two or three tests. However, the relationship between all tests conducted prior to the specific test and the specific test used in the prediction model creation process, and the format of the interval between the number of tests in the dataset used in the prediction processing process described below, including tests conducted after the latest of the specific tests, and the test to be predicted, must be the same. Specifically, the number of tests between the specific test and the target test (how many tests were conducted before the specific test and the target test) and the number of tests going back from there (how many consecutive tests) selected as explanatory variables must be the same. For example, a dataset with a specific test period set to FY22 and multiple tests set to FY2020-2011, a dataset with a specific test period set to FY21 and multiple tests set to FY2019-2010, and a dataset with a specific test period set to FY2020 and multiple tests set to FY2018-2009 may be used. The number of tests to go back to the specific test period and the number of tests to be selected from that number shall be consistent across all datasets. Furthermore, in this case, when selecting FY23 as the test period to be predicted, the number of tests to go back to the specific test period and the number of tests to be selected from that number shall be consistent with the dataset for the specific test period, such as entering FY21-2012 as explanatory variables. Note that this data consistency only requires that the data used as input for the prediction model creation step S2 and the prediction processing step S3 be consistent. The number of data (number of times) for each data set does not need to be consistent due to the inclusion of other data or other data not used as input explanatory variables.

[0063] (Step S3: Prediction processing step) Step S3 is a prediction processing step, which predicts the possibility of a question being asked in the exam being predicted, or calculates an index related to the possibility of a question being asked, for each correct answer question, based on whether or not the question has been asked in multiple exams, including the exam after the latest of the multiple exams held before the specific exam, contained in the question data, and a prediction model.

[0064] To explain this using the specific example shown in step S2, if the exam to be predicted is for the 2023 exam, for example, for each correct answer point, the exams from 2021 to 2012 are entered as explanatory variables into the prediction model created as described above.

[0065] (When adding the classification of the correct answer's argument as an explanatory variable) In step S2, if a prediction model is to be created that takes into consideration the classification ID, the prediction model may be created by further adding the classification ID as an explanatory variable.

[0066] Although not shown, a display control step and a display step may be provided. The display control step is a step of controlling the display of the probability of a question being asked in a subsequent exam, predicted in step S3, or an index related to the probability of the question being asked calculated in step S3, on a display device such as a display. The display step is a step of displaying the probability of the question being asked or an index related thereto on a display device such as a display.

[0067] The prediction device, prediction program, and prediction method described above can be used for any exam, including qualification exams, entrance exams, and the like, as long as the exam includes multiple-choice questions.

[0068] The prediction device, prediction program, and prediction method described above can be implemented with appropriate modifications as long as the effects of the present invention are not impaired. [Explanation of symbols]

[0069] 11 Question Data Reception Department 12 Prediction model memory section 13 Prediction processing section 14 Display control unit 15 Display 16 Memory section 17 Communications Department 18 Control Unit 19 Communication Bus

Claims

1. A prediction device that predicts the likelihood of multiple-choice questions being asked in a future target test based on question trends in past tests, a question data receiving unit that receives question data including whether or not a question on each of the points of the correct answer options that are correct answers to the multiple-choice questions in the past multiple exams has been asked in each of the past consecutive exams prior to the exam to be predicted; a prediction model creation unit that creates, for each point of the correct answer, a prediction model that indicates the relationship between the presence or absence of a question in multiple exams taken before the target time to be predicted and the possibility of the question being asked in the target time to be predicted and / or an index related to the possibility of the question being asked, based on multiple data sets of whether the question was asked in multiple exams taken before the specific time and whether the question was asked in the specific time, which are included in the question data; a prediction processing unit that predicts the possibility of the question being asked in the exam to be predicted, or calculates an index related to the possibility of the question being asked, based on whether or not the question has been asked in multiple exams, including the exam after the latest of the multiple exams held before the specific exam, which are included in the question data, and the prediction model, for each point of the correct answer; A prediction device comprising:

2. The prediction model is a logistic regression model in which the presence or absence of a question in a plurality of exams conducted prior to the prediction target exam is used as an explanatory variable, The prediction processing unit calculates the probability that the correct answer point will be included in the question. The prediction device according to claim 1 .

3. The prediction model is a logistic regression model in which the presence or absence of a question in a plurality of exams conducted prior to the prediction target exam is used as an explanatory variable, The prediction processing unit determines the learning priority of the point of the correct answer. The prediction device according to claim 1 .

4. A prediction program that predicts the likelihood of multiple-choice questions being asked in a future target test based on question trends in past tests, The computer is caused to function as a question data receiving unit, a prediction model creating unit, and a prediction processing unit; the question data receiving unit receives, for each point of a correct answer that is a correct answer to the multiple-choice question in the past multiple exams, question data including whether or not the point has been asked in each of the past consecutive exams prior to the prediction target exam; The prediction model creation unit creates a prediction model showing the relationship between the presence or absence of a question in a plurality of exams conducted before the prediction target exam and the possibility of the question being asked in the prediction target exam and / or an index related to the possibility of the question being asked, based on a plurality of data sets of whether the question was asked in a plurality of exams conducted before a specific exam and whether the question was asked in the specific exam, which are included in the question data, for each point of the correct answer; The prediction processing unit predicts the possibility of the question being asked in the exam to be predicted, or calculates an index related to the possibility of the question being asked, based on whether the question has been asked in a plurality of exams, including an exam after the latest of the plurality of exams held before the specific exam, which are included in the question data, and the prediction model, for each point of the correct answer; Prediction program.

5. A prediction method for predicting the likelihood of multiple-choice questions being asked in a future target test based on question trends in past tests, a question data receiving step for receiving question data including whether or not a question on each of the points of the correct answer options that are correct answers to the multiple-choice questions in the past multiple examinations has been asked in each of the past consecutive examinations prior to the examination to be predicted; a prediction model creation process for creating a prediction model that indicates the relationship between the presence or absence of a question in a plurality of exams conducted prior to the target test and the likelihood of the question being asked in the target test and / or an index related to the likelihood of the question being asked, based on multiple data sets of whether the question was asked in a plurality of exams conducted prior to the target test and whether the question was asked in the target test, for each of the points of the correct answer; a prediction processing step of predicting the possibility of the question being asked in the exam to be predicted, or calculating an index related to the possibility of the question being asked, based on whether or not the question has been asked in a plurality of exams, including an exam after the latest of a plurality of exams held before the specific exam, which are included in the question data, and the prediction model, for each point of the correct answer; A forecasting method including:

Images