Information processing device, information processing method, and program

Abstract

[Problem] To predict, with high accuracy, a behavior that potentially would be executed in a predetermined condition by a person to be analyzed. [Solution] Provided is an information processing device provided with a prediction unit for, on the basis of preference past record data indicating the past record of a preference behavior related to a predetermined task executed by a person to be analyzed, outputting preference prediction data indicating prediction of the preference behavior that can be executed by the person to be analyzed in a predetermined condition. The prediction unit inputs the preference past record data to a classifier generated by manifold learning, and outputs the preference prediction data on the basis of applying of a prediction model based on assumed information related to the predetermined condition for each of a plurality of units obtained through classification.

Description

[Technical Field]

[0001] The present disclosure relates to an information processing device, an information processing method, and a program. [Background technology]

[0002] It is extremely important to properly analyze the behavior of a task performer who performs a certain task. For this reason, in recent years, many mechanisms have been proposed to automate or assist such analysis. For example, Patent Document 1 proposes a mechanism that analyzes the investment behavior of an investment trust fund (hereinafter simply referred to as a fund) and provides a rating based on the results of the analysis. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2009-245368 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the rating method disclosed in Patent Document 1 is based on past performance and is not sufficient for verifying the future usefulness of a fund. [Means for solving the problem]

[0005] According to one aspect of the present disclosure, there is provided an information processing device including a prediction unit that outputs preference prediction data indicating a prediction of preference behavior that may be performed by an analyzable person in a specified situation based on preference performance data indicating the performance of preference behavior related to a specified task performed by the analyzable person, wherein the prediction unit inputs the preference performance data to a classifier generated by manifold learning, and outputs the preference prediction data based on applying a prediction model based on assumed information related to the specified situation to each of a plurality of classified units.

[0006] According to another aspect of the present disclosure, there is provided an information processing method including a processor outputting preference prediction data indicating a prediction of preference behavior that may be performed by an analyzable person in a predetermined situation based on preference performance data indicating the performance of preference behavior related to a predetermined task performed by the analyzable person, wherein the outputting further includes inputting the preference performance data to a classifier generated by manifold learning, and outputting the preference prediction data based on applying a prediction model based on assumed information related to the predetermined situation to each of a plurality of classified units.

[0007] According to another aspect of the present disclosure, there is provided a program for causing a computer to function as an information processing device, comprising: a prediction unit that outputs preference prediction data indicating a prediction of preference behavior that may be performed by an analyzable person in a specified situation based on preference performance data indicating the performance of preference behavior related to a specified task performed by the analyzable person, wherein the prediction unit inputs the preference performance data to a classifier generated by manifold learning, and outputs the preference prediction data based on applying a prediction model based on assumed information related to the specified situation to each of a plurality of classified units. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a block diagram illustrating an example of a functional configuration of a learning device 10 according to an embodiment of the present disclosure. [Figure 2] FIG. 2 is a block diagram showing an example of a functional configuration of a prediction device 20 according to the embodiment. [Figure 3] FIG. 10 is a diagram for explaining the classification of preference objects using a self-organizing map by the prediction unit 210 according to the embodiment, and the generation of a map that represents the strength of indices based on the preference ratios of the classified preference objects in the form of a heat map. [Figure 4] FIG. 10 is a schematic diagram for explaining output of preference prediction data by the prediction unit 210 according to the embodiment. [Figure 5]FIG. 10 is a diagram showing an example of a map that represents the strength of predicted preference ratios of preferred objects in a heat map format, generated by the prediction unit 210 according to the embodiment. [Figure 6] 10 is an example of a map generated by the prediction unit 210 according to the embodiment, which represents, in a heat map format, the strength of the difference between predicted information of active weights in a predetermined situation and actual information of active weights in the past for each BMU. [Figure 7] FIG. 10 is a diagram showing an example of a map based on preference prediction data showing predictions of preference behaviors that may be performed by multiple subjects according to the embodiment. [Figure 8] 10 is a flowchart showing an example of a flow of prediction of preference behavior by the prediction unit 210 according to the embodiment. [Figure 9] FIG. 2 is a block diagram showing an example of the hardware configuration of an information processing device 90 according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] Preferred embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. In this specification and drawings, components having substantially the same functional configurations are designated by the same reference numerals, and redundant description will be omitted.

[0010] The explanation will be given in the following order. 1. Embodiment Overview 1.2. System configuration example 1.3. Forecast Details 1.4.Prediction process 2. Hardware configuration example 3. Summary

[0011] <1. Embodiment> <<1.1. Overview>> For example, when an organization selects a new task performer to be entrusted with a task, it is important to accurately predict how the task performer will act in various situations that may arise in the future.

[0012] Furthermore, if it were possible to accurately predict the actions that a task performer might take, it would theoretically be possible to obtain the same benefits by having other personnel or a system perform the predicted actions instead of hiring a new task performer.

[0013] However, the more complex the task, the more difficult it becomes to predict the actions that a task performer may take.

[0014] As an example, assume that the task is asset management and the task performer is a fund that invests in financial products.

[0015] Generally, investment activities by funds are considered to require highly specialized knowledge and are carried out based on complex decision-making.

[0016] For this reason, it is believed to be difficult to accurately predict the investment behavior of funds and to replicate (imitate) the predicted results.

[0017] The technical concept of one embodiment of the present disclosure was conceived with the above points in mind, and makes it possible to accurately predict the actions that may be performed by a task performer (analyzed person) in a given situation.

[0018] For this reason, one of the features of the information processing method according to this embodiment is that it utilizes the property of a manifold that analyzes the object of observation into a region that can be locally linearized, and converts seemingly complex behavioral patterns into response functions.

[0019] That is, in the information processing method according to this embodiment, behavioral patterns that appear to be complex are divided into modelable scales, and a response function is generated for each scale.

[0020] For this purpose, the prediction device 20 that executes the information processing method of this embodiment includes a prediction unit 210 that outputs preference prediction data that indicates a prediction of preference behavior that may be performed by the person being analyzed in a specified situation, based on preference performance data that indicates the performance of preference behavior related to a specified task performed by the person being analyzed.

[0021] Furthermore, one of the features of the prediction unit 210 according to this embodiment is that it inputs preference performance data into a classifier generated by manifold learning, and outputs preference prediction data based on applying a prediction model based on assumed information relating to a predetermined situation to each of the classified units.

[0022] In the following, a case will be mainly described as an example in which the predetermined task is asset management and the preference behavior is investment behavior in financial products.

[0023] In this case, the preference history data may include information on past active weights based on investment behaviors performed by the subject of analysis, and the preference prediction data may include information on predictions of active weights based on investment behaviors that may be performed in a given situation.

[0024] That is, the prediction device 20 according to this embodiment may predict an active weight determined by the investment behavior that the fund selected as the subject of analysis may perform in a certain situation, based on the active weight determined by the investment behavior that the fund has performed in the past.

[0025] For example, when trying to predict profits (fund returns) when a certain fund is adopted, one possible method is to first predict factor returns and then predict fund returns based on the predicted factor returns.

[0026] However, information such as factor returns is heavily influenced by uncontrollable market fluctuations and fluctuates independently of fund decision-making.

[0027] In other words, since information such as factor returns is heavily dependent on external information whose occurrence is random and whose strength is unpredictable, it can be said that forecasts of fund returns based on forecasts of factor returns are highly random and unreliable.

[0028] On the other hand, active weighting is the result of the fund's investment behavior and does not fluctuate independently of the fund's decision-making.

[0029] Active weighting is also affected by external information, which occurs randomly and is unpredictable in intensity. However, since funds do not react sensitively to all external information and are expected to make long-term investments, it can be said that it is less affected by external information than factor returns, etc.

[0030] As a result, when predicting active weights as a result of investment actions that a fund may take in the future, as in the information processing method of this embodiment, it is possible to achieve highly accurate predictions that are closely related to the fund's decision-making.

[0031] It is assumed that the information processing method according to this embodiment is particularly effective in the use cases shown below, for example.

[0032] As an example, the information processing method according to this embodiment can be applied to the refinement of risk scenario analysis.

[0033] In a typical risk scenario analysis, the current portfolio is treated as fixed and losses are estimated in the event of a market risk occurring.

[0034] On the other hand, the information processing method according to this embodiment is based on the premise that the portfolio changes dynamically depending on the market conditions, which makes it possible to estimate losses more precisely by taking into account the investment behavior of the fund (analyzed party) as a response to market risks, etc.

[0035] Furthermore, the information processing method according to this embodiment makes it possible to predict the investment behavior of all entrusted investment managers (task performers) in response to any market change according to their respective characteristics, and to estimate which funds are likely to behave similarly, etc. This makes it possible to quantitatively evaluate the possibility that the manager structure (the composition of investment managers to be employed as entrusted managers) expected in advance will change.

[0036] Furthermore, as an example, the information processing method according to this embodiment can be applied to the selection of an investment management institution, for example.

[0037] For example, when an institution seeks to recruit a new management institution, the information available to the institution on candidate management institutions is limited compared to the information available to management institutions that have already signed contracts.

[0038] On the other hand, the information processing method according to this embodiment makes it possible to predict in advance the investment behavior of candidate investment managers in various market environments by learning from the investment behavior of candidate investment managers. This makes it possible to build a new, highly effective manager structure.

[0039] Moreover, as an example, the information processing method according to this embodiment can be applied to support dialogue with funds.

[0040] Generally, asset managers at funds have a high level of investment expertise. However, if the person in charge at an institution does not have the same level of expertise as the asset manager, the person in charge may be unable to object to the statements made by the asset manager and may be forced to accept the statements at face value.

[0041] However, the information processing method according to this embodiment makes it possible to predict the investment behavior of the analyzed person in advance. This allows the person in charge at the institution to understand changes in style related to investment behavior and abnormal trades by comparing the prediction with the actual results, and enables dialogue at the same level as the asset manager.

[0042] Moreover, as an example, the information processing method according to this embodiment can be applied to fund duplication.

[0043] As described above, the information processing method according to this embodiment makes it possible to model the investment behavior of a fund selected as a target for analysis based on the past performance of that investment behavior. This makes it possible to incorporate the investment behavior predicted by the model into in-house management, thereby introducing the target's sophisticated strategy at low cost.

[0044] <<1.2. System configuration example>> Next, an example of a system configuration according to this embodiment will be described in detail. The system according to this embodiment includes a learning device 10 that performs manifold learning using a machine learning algorithm, and a prediction device 20 that performs prediction using a classifier generated by the manifold learning performed by the learning device 10.

[0045] (Learning Device 10) First, a description will be given of an example of the functional configuration of a learning device 10 according to this embodiment. Fig. 1 is a block diagram showing an example of the functional configuration of a learning device 10 according to this embodiment.

[0046] As shown in FIG. 1, the learning device 10 according to this embodiment may include a learning unit 110 and a storage unit 120.

[0047] (Learning Section 110) The learning unit 110 according to this embodiment performs manifold learning using a machine learning algorithm.

[0048] For example, the learning section 110 according to this embodiment learns a classification related to preference performance data based on preference performance data indicating performance of preference behaviors related to a predetermined task performed by the subject.

[0049] The preference performance data according to this embodiment may include situation transition data indicating the transition of past situations, and past preference ratio data indicating the preference ratio of preference objects that were the subject of preference behavior in the past situations.

[0050] As described above, the preference behavior may be investment behavior in which a financial product (for example, a brand) to invest in and an investment amount are selected from a plurality of financial products.

[0051] In this case, the preference performance data according to this embodiment can be said to be investment performance data that indicates the performance of the investment behavior performed by the person to be analyzed.

[0052] In this case, the situation transition data included in the preference performance data may be data indicating past transitions in the market environment.

[0053] Examples of the situation transition data include factor returns and factor properties.

[0054] As the factor return, the market return, the return difference between value and growth, the return difference between small and large, momentum, or the like may be adopted.

[0055] Furthermore, excess return over a benchmark, market capitalization, price-to-book ratio (PBR), or the like may be adopted as the factor property.

[0056] Furthermore, the past preference ratio data included in the preference performance data may be performance information of past active weights that indicates the investment ratio of stocks that could have been the subject of investment behavior.

[0057] The learning unit 110 according to this embodiment may input the above-mentioned preference performance data into a neural network and perform manifold learning to classify stocks into a plurality of units (BMUs: Best Matching Units).

[0058] An example of the above manifold learning is a self-organizing map (SOM).

[0059] The functions of the learning unit 110 according to this embodiment are realized by a processor such as a GPU.

[0060] (Storage unit 120) The storage unit 120 according to this embodiment stores various types of information related to manifold learning executed by the learning unit 110. For example, the storage unit 120 stores the structure of a network used in manifold learning by the learning unit 110, various parameters related to the network, learning data, and the like.

[0061] The above describes an example of the functional configuration of the learning device 10 according to this embodiment. Note that the functional configuration described above using Fig. 1 is merely an example, and the functional configuration of the learning device 10 according to this embodiment is not limited to this example.

[0062] For example, the study device 10 according to this embodiment may further include an operation unit that accepts operations by the user, a display unit that displays various types of information, and the like.

[0063] The functional configuration of the learning device 10 according to this embodiment can be flexibly modified according to specifications and operations.

[0064] (Prediction device 20) Next, a description will be given of an example of the functional configuration of the prediction device 20 according to this embodiment. The prediction device 20 according to this embodiment is an example of an information processing device that performs prediction using a classifier generated by manifold learning by the learning device 10.

[0065] 2 is a block diagram showing an example of the functional configuration of the prediction device 20 according to this embodiment. As shown in FIG. 2, the prediction device 20 according to this embodiment may include a prediction unit 210, a storage unit 220, a display unit 230, and an operation unit 240.

[0066] (Prediction unit 210) The prediction unit 210 according to this embodiment outputs preference prediction data indicating a prediction of preference behavior that may be performed by the subject in a specified situation, based on preference performance data indicating the performance of preference behavior related to a specified task performed by the subject.

[0067] Furthermore, one of the features of the prediction unit 210 according to this embodiment is that it inputs actual preference data into a classifier generated by manifold learning using the learning device 10, and outputs preference prediction data based on applying a prediction model based on assumed information relating to a predetermined situation to each of the classified units.

[0068] As described above, the classifier generated by manifold learning by the learning device 10 may be a self-organizing map.

[0069] The prediction section 210 according to this embodiment accurately predicts actions that may be performed by the analyte task in a given situation.

[0070] The functions of the prediction unit 210 according to this embodiment will be described in detail separately. The functions of the prediction unit 210 according to this embodiment are realized by a processor such as a GPU.

[0071] (Storage unit 220) The storage unit 220 according to this embodiment stores various types of information used by the prediction device 20. The storage unit 220 stores, for example, preference achievement data, the structure and parameters of a classifier used by the prediction unit 210, preference prediction data output by the prediction unit 210, and the like.

[0072] (Display section 230) The display unit 230 according to the present embodiment displays various types of visual information, and therefore includes a display.

[0073] For example, the display unit 230 according to this embodiment displays the results of prediction by the prediction unit 210 under the control of the prediction unit 210. The results of prediction include various maps generated by the prediction unit 210.

[0074] (Operation unit 240) The operation unit 240 according to this embodiment accepts operations by the user, and therefore includes various input devices such as a keyboard and a mouse.

[0075] The functional configuration of the prediction device 20 according to this embodiment has been described above. Note that the functional configuration described above with reference to Fig. 2 is merely an example, and the functional configuration of the prediction device 20 according to this embodiment is not limited to this example.

[0076] For example, the prediction unit 210 and the storage unit 220 according to this embodiment, and the display unit 230 and the operation unit 240 may be provided in separate devices. For example, the prediction unit 210 and the storage unit 220 may be provided in an information processing device located on the cloud, and the display unit 230 and the operation unit 240 may be provided in an information processing device located locally.

[0077] The functional configuration of the prediction device 20 according to this embodiment can be flexibly modified according to specifications and operations.

[0078] <<1.3. Forecast Details>> Next, prediction by the prediction unit 210 according to this embodiment will be described in detail.

[0079] As described above, the prediction unit 210 according to this embodiment outputs preference prediction data indicating a prediction of preference behavior that may be performed by the subject in a specified situation, based on preference performance data indicating the performance of preference behavior related to a specified task performed by the subject.

[0080] In the following, a case will be described in which the predetermined task is asset management and the preference behavior is investment behavior in financial products.

[0081] In this case, the preference performance data according to this embodiment may be investment performance data indicating the performance of investment behaviors carried out by the person to be analyzed.

[0082] Furthermore, the situation transition data included in the preference performance data may be data indicating past transitions in the market environment.

[0083] Furthermore, the past preference ratio data included in the preference performance data may be performance information of past active weights that indicate the investment ratio of stocks that could have been the subject of investment behavior.

[0084] The prediction unit 210 of this embodiment can classify multiple preference targets, i.e., multiple stocks, into multiple specified BMUs (hereinafter, sometimes simply referred to as units) by inputting the above-mentioned preference performance data into a self-organizing map.

[0085] In addition, at this time, the prediction unit 210 according to this embodiment may generate a map that expresses, in the form of a heat map, the strength of the index based on the preference ratio of the stocks belonging to each unit.

[0086] FIG. 3 is a diagram for explaining the classification of preference objects using a self-organizing map by the prediction unit 210 according to this embodiment, and the generation of a map that represents the strength of indices based on the preference ratios of the classified preference objects in the form of a heat map.

[0087] For example, the left side of Figure 3 shows maps M1 and M3 generated by the prediction unit 210 based on preference performance data for Fund A, which is the subject of analysis. The right side of Figure 3 shows maps M2 and M4 generated by the prediction unit 210 based on preference performance data for Fund B, which is the subject of analysis.

[0088] Here, the preference performance data for Fund A and the preference performance data for Fund B are acquired during the same period, and the status transition data included in both preference performance data may be the same.

[0089] In this case, the preferred stocks can be classified into the same units in the maps M1 to M4.

[0090] On the other hand, the past preference ratio data (past active weight performance information) included in the preference performance data for Fund A and the past preference ratio data included in the preference performance data for Fund B are different from each other.

[0091] For this reason, the prediction unit 210 may generate maps M1 and M2 that represent the strength of preference ratios of stocks classified by unit in the form of a heat map based on past performance information of active weights, as shown in the upper part of Figure 3.

[0092] Furthermore, as shown in the lower part of Figure 3, the prediction unit 210 may generate maps M3 and M4 that represent the strength of the excess returns over the benchmark of stocks classified by unit in the form of a heat map, based on past performance information of active weights and the excess returns over the benchmark for each stock.

[0093] In the example shown in Figure 3, the intensity of each index is expressed using dots and diagonal lines. Specifically, when a unit is expressed by dots, the intensity is low, and the higher the density of the dots, the lower the degree of intensity. On the other hand, when a unit is expressed by diagonal lines, the intensity is high, and the higher the density of the diagonal lines, the higher the degree of intensity.

[0094] Comparing Map M1 and Map M2, for example, it can be seen that Fund A holds more stocks classified in units located at the bottom center of the map than the market average, while Fund B holds more stocks classified in units located at the top right of the map than the market average.

[0095] Furthermore, comparing maps M1 and M3, it can be seen that Fund A holds more stocks classified as units located in the lower center of the map than the market average, and thereby earns higher returns than the market average for those stocks.

[0096] On the other hand, comparing maps M2 and M4, it can be seen that Fund B holds more stocks classified as units located on the right side of the map than the market average, and thereby earns higher returns than the market average for those stocks.

[0097] Furthermore, comparing maps M2 and M4, it can be seen that Fund B is earning higher returns than the market average for stocks classified as units located slightly below the center on the right side of the map by holding fewer stocks than the market average.

[0098] In this way, according to the prediction unit 210 of this embodiment, by performing classification using a self-organizing map and creating a heat map based on the attributes of each stock, it becomes possible to analyze the differences and similarities in the investment behavior of each fund.

[0099] Furthermore, the prediction unit 210 according to this embodiment can output preference prediction data by applying a prediction model based on assumed information relating to a predetermined assumed situation for each unit.

[0100] FIG. 4 is a schematic diagram for explaining the output of preference prediction data by the prediction unit 210 according to this embodiment.

[0101] As described above, the preference history data input to the self-organizing map according to this embodiment includes past preference ratio data indicating the preference ratio of preference objects that could have been the subject of preference behavior in past situations.

[0102] When the preference performance data is investment performance data by the subject of analysis, the above-mentioned past preference ratio data may be performance information of active weights in the past.

[0103] In this case, the prediction unit 210 according to this embodiment inputs the preference performance data into a self-organizing map, classifying the preference targets, i.e., multiple stocks that could be investment targets, into multiple BMUs, and obtains a first codebook vector for each BMU.

[0104] At this time, in the self-organizing map according to this embodiment, the preference performance data including the performance information of the active weights is subjected to standardization and normalization processing that can be reversed, and each stock is classified into a plurality of units.

[0105] Here, the above-mentioned first codebook vector can be said to be a variable obtained for each BMU (0 to n) and corresponding to the preference ratio (investment ratio) of the stocks classified into each BMU.

[0106] Next, the prediction unit 210 according to this embodiment obtains a second codebook vector for each BMU by applying a prediction model based on assumed information relating to a specific assumed situation to the first codebook vector obtained for each BMU.

[0107] Here, the second codebook vector can be said to be a variable obtained for each BMU (0 to n) and corresponding to the predicted preference ratio (predicted investment ratio) in a predetermined situation of the stocks classified into each BMU.

[0108] The above-mentioned assumed information may include at least one of assumed information on factor returns or assumed information on factor properties in a predetermined situation assumed by the analyst.

[0109] For example, expected information on factor returns may include expected information on market returns in a given assumed situation, the difference in returns between value and growth, the difference in returns between small and large stocks, or momentum.

[0110] On the other hand, assumed information for factor properties may include excess return over a benchmark, market capitalization, or price-to-book ratio in a given assumed situation.

[0111] The analyst may imagine any situation in which he / she wishes to predict the preferred behavior of the person being analyzed, and set assumed information related to that situation.

[0112] The above-mentioned predetermined situation may be, for example, the market environment several months or a year from now if there are no major changes, or the market environment several months or a year from now if the yen suddenly appreciates.

[0113] Furthermore, examples of prediction models based on the above-mentioned assumed information include multiple regression models, vector autoregression models, and GNN (Graphical Neural Network) models.

[0114] The prediction model according to this embodiment can be set appropriately based on the tendency of the analysis subject's preference behavior.

[0115] For example, a multiple regression model may be used to predict the preference behavior of discretionary subjects who trade relatively infrequently.

[0116] Here, when the factor return in a given assumed situation (t) is expressed as the following mathematical formula (1), the second codebook vector CV obtained by the multiple regression model is expressed, for example, by the following mathematical formula (2). Note that B in mathematical formula (2) is the regression coefficient for each factor return estimated from past FR and CV.

[0117]

number

[0118] On the other hand, for example, a vector autoregression model may be used to predict the preference behavior of a Quants-type subject who trades relatively frequently or a subject who changes style significantly.

[0119] The second codebook vector CV obtained by the vector autoregressive model is expressed by, for example, the following formula (3). (i) indicates the vector corresponding to the i-th BMU among the regression coefficients B estimated as above, and B t is expressed by the following formula (4): F and Q in formula (4) are matrices that represent the characteristics of the entire analyzed fund when viewed as a single system, estimated from past CV trends.

[0120]

number

[0121] On the other hand, for example, a GNN-type model may be used when simultaneously processing predictions of preference behavior by multiple types of subjects (for example, predictions of preference behavior by manager structure).

[0122] Next, the prediction unit 210 according to this embodiment performs standardization and inverse normalization transformation on the second codebook vector acquired for each BMU using the above-described prediction model, and outputs preference prediction data.

[0123] The preference prediction data may include predicted preference ratio data indicating predicted preference ratios of preferred objects in a given situation.

[0124] For example, if the preference performance data used as input to the self-organizing map is investment performance data by the subject of analysis, the predicted preference ratio data may be predicted information on active weights indicating the holding ratio of each stock under specified circumstances.

[0125] The output of predicted change ratio data by the prediction section 210 according to this embodiment has been described in detail above.

[0126] The prediction unit 210 according to this embodiment may generate a map that represents the strength of the predicted preference ratio of the preferred object for each BMU in the form of a heat map, based on the predicted change ratio data output as described above.

[0127] FIG. 5 is a diagram showing an example of a map in which the strength of the predicted preference ratio of a preferred object is expressed in the form of a heat map, generated by the prediction unit 210 according to this embodiment.

[0128] The left side of Figure 5 shows a map M5 generated by the forecasting unit 210 based on the preference performance data and predicted preference ratio data for Fund A, which is the subject of analysis. The right side of Figure 5 shows a map M6 generated by the forecasting unit 210 based on the preference performance data and predicted preference ratio data for Fund B, which is the subject of analysis.

[0129] The prediction unit 210 according to this embodiment can generate maps M5 and M6 that represent the strength of the predicted holding ratio in a specified situation for each stock classified by BMU in the form of a heat map based on the output predicted change ratio data, i.e., the predicted information of the active weight in a specified situation.

[0130] For example, by comparing map M1 shown in Figure 3 with map M5 shown in Figure 5, an analyst can visually and intuitively understand how the active weight for Fund A will change under certain circumstances.

[0131] Similarly, by comparing map M2 shown in Figure 3 with map M6 shown in Figure 5, an analyst can visually and intuitively understand how the active weight for Fund B will change under certain circumstances.

[0132] For the above-described comparison, the prediction unit 210 according to this embodiment may control the display unit 230 to display the maps M1 and M5, and the maps M2 and M6 side by side.

[0133] On the other hand, the prediction unit 210 according to this embodiment may generate a map that represents the magnitude of the difference between the predicted preference ratio data and the past preference ratio data for each BMU in the form of a heat map, and display the map on the display unit 230.

[0134] For example, the prediction unit 210 according to this embodiment may generate a map M7 that represents the magnitude of the difference between the predicted information of the active weight in a given situation and the actual information of the active weight in the past in the form of a heat map for each BMU, as shown in FIG.

[0135] In map M7, the magnitude of the difference between the predicted information of the active weight in a given situation and the actual information of the active weight in the past is expressed using dots and diagonal lines. Specifically, when a BMU is expressed by a dot, the difference is small, and the higher the density of the dots, the smaller the difference is. On the other hand, when a BMU is expressed by a diagonal line, the difference is large, and the higher the density of the diagonal lines, the larger the difference is.

[0136] According to the map described above, the analyst can intuitively grasp how the active weight for the person being analyzed changes in a given situation.

[0137] 5 and 6, the prediction unit 210 outputs preference prediction data relating to a single subject to be analyzed, and generates a map based on the preference prediction data.

[0138] On the other hand, the prediction unit 210 according to this embodiment may output preference prediction data indicating a prediction of preference behaviors that may be performed by the multiple subjects in a specified situation, based on multiple preference actual data relating to the multiple subjects.

[0139] That is, the prediction unit 210 according to this embodiment can predict preference behavior for each manager structure.

[0140] FIG. 7 is a diagram showing an example of a map based on preference prediction data showing predictions of preference behaviors that may be performed by a plurality of subjects according to this embodiment.

[0141] For example, map M8 shown in Figure 7 is an example of a map generated by the prediction unit 210 based on preference prediction data obtained by inputting data that is a 1:1 merge of preference performance data for fund A and preference performance data for fund C into a self-organizing map.

[0142] According to the map M8 as described above, an analyst can visually and intuitively grasp what active weights will be formed in a given situation if an equal amount of assets are allocated to Fund A and Fund C.

[0143] On the other hand, map M9 shown in Figure 7 is an example of a map generated by the prediction unit 210 based on preference prediction data obtained by inputting data obtained by merging preference performance data related to fund B and preference performance data related to fund C in a 3:1 ratio into a self-organizing map.

[0144] According to the map M9 shown above, an analyst can visually and intuitively grasp what active weights will be formed in a given situation if Fund B and Fund C are adopted and three times as many assets as Fund C are allocated to Fund B.

[0145] As described above, the prediction unit 210 according to this embodiment can accurately predict the actions that may be performed by one or more subjects in a given situation, and can also visualize the prediction results.

[0146] <<1.4. Prediction process>> Next, an example of the flow of predicting preference behavior by the predicting unit 210 according to this embodiment will be described in detail.

[0147] FIG. 8 is a flowchart showing an example of the flow of prediction of preference behavior by the prediction unit 210 according to this embodiment.

[0148] In the example shown in FIG. 8, the prediction unit 210 first inputs preference performance data into the self-organizing map (S102).

[0149] Next, the prediction unit 210 applies a prediction model based on the assumption information for each BMU to obtain a second codebook vector (S104).

[0150] Next, the prediction unit 210 performs standardization and inverse normalization on the second codebook vector acquired in step S104, and outputs preference prediction data (S106).

[0151] Next, the prediction unit 210 generates various maps based on the preference prediction data output in step S106 (S108).

[0152] <2. Hardware configuration example> Next, an example of a hardware configuration common to the learning device 10 and the prediction device 20 according to an embodiment of the present disclosure will be described. Fig. 9 is a block diagram showing an example of a hardware configuration of an information processing device 90 according to an embodiment of the present disclosure. The information processing device 90 may be a device having a hardware configuration equivalent to that of each of the above devices.

[0153] 9, the information processing device 90 includes, for example, a processor 871, a ROM 872, a RAM 873, a host bus 874, a bridge 875, an external bus 876, an interface 877, an input device 878, an output device 879, a storage 880, a drive 881, a connection port 882, and a communication device 883. Note that the hardware configuration shown here is an example, and some of the components may be omitted. Furthermore, the information processing device 90 may further include components other than those shown here.

[0154] (Processor 871) The processor 871 functions, for example, as an arithmetic processing device or control device, and controls the overall operation of each component or part of it based on various programs recorded in the ROM 872, RAM 873, storage 880, or removable storage medium 901.

[0155] (ROM872, RAM873) The ROM 872 is a means for storing programs to be read into the processor 871, data to be used for calculations, etc. The RAM 873 temporarily or permanently stores, for example, programs to be read into the processor 871, and various parameters that change as appropriate when the programs are executed.

[0156] (Host bus 874, bridge 875, external bus 876, interface 877) The processor 871, ROM 872, and RAM 873 are connected to one another via, for example, a host bus 874 that is capable of high-speed data transmission. On the other hand, the host bus 874 is connected to, for example, an external bus 876 that has a relatively low data transmission speed via a bridge 875. In addition, the external bus 876 is connected to various components via an interface 877.

[0157] (Input Device 878) The input device 878 may be, for example, a mouse, keyboard, touch panel, button, switch, lever, etc. Furthermore, a remote controller (hereinafter referred to as a remote control) capable of transmitting control signals using infrared rays or other radio waves may also be used as the input device 878. The input device 878 may also include an audio input device such as a microphone.

[0158] (Output Device 879) The output device 879 is a device capable of visually or audibly notifying the user of acquired information, such as a display device such as a CRT (Cathode Ray Tube), LCD, or organic EL, an audio output device such as a speaker or headphones, a printer, a mobile phone, a facsimile, etc. The output device 879 according to the present disclosure also includes various vibration devices capable of outputting tactile stimuli.

[0159] (Storage 880) The storage 880 is a device for storing various types of data. For example, a magnetic storage device such as a hard disk drive (HDD), a semiconductor storage device, an optical storage device, or a magneto-optical storage device may be used as the storage 880.

[0160] (Drive 881) The drive 881 is a device that reads information recorded on a removable storage medium 901 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, or writes information to the removable storage medium 901 .

[0161] (Removable storage medium 901) The removable storage medium 901 is, for example, a DVD medium, a Blu-ray (registered trademark) medium, an HD DVD medium, various semiconductor storage media, etc. Of course, the removable storage medium 901 may also be, for example, an IC card equipped with a contactless IC chip, an electronic device, etc.

[0162] (Connection port 882) The connection port 882 is a port for connecting an external device 902, such as a USB (Universal Serial Bus) port, an IEEE1394 port, a SCSI (Small Computer System Interface), an RS-232C port, or an optical audio terminal.

[0163] (External connection device 902) The externally connected device 902 is, for example, a printer, a portable music player, a digital camera, a digital video camera, or an IC recorder.

[0164] (Communication Device 883) The communication device 883 is a communication device for connecting to a network, such as a communication card for wired or wireless LAN, Bluetooth (registered trademark), or WUSB (Wireless USB), a router for optical communication, a router for ADSL (Asymmetric Digital Subscriber Line), or a modem for various types of communication.

[0165] <3. Summary> As described above, the prediction unit 210 according to one embodiment of the present disclosure outputs preference prediction data indicating a prediction of preference behavior that may be performed by the subject in a specified situation, based on preference performance data indicating the performance of preference behavior related to a specified task performed by the subject.

[0166] Furthermore, one of the features of the prediction unit 210 according to an embodiment of the present disclosure is that it inputs preference performance data into a classifier generated by manifold learning using the learning device 10, and outputs preference prediction data based on applying a prediction model based on assumed information relating to a predetermined situation to each of the classified units.

[0167] According to the above configuration, it is possible to accurately predict the actions that the subject may take in a given situation.

[0168] Although the preferred embodiments of the present disclosure have been described in detail above with reference to the accompanying drawings, the technical scope of the present disclosure is not limited to such examples. It is clear that a person skilled in the art of the present disclosure can conceive of various modified or altered examples within the scope of the technical idea described in the claims, and it is understood that these also naturally fall within the technical scope of the present disclosure.

[0169] For example, in the above embodiment, the predetermined task is asset management, and the analysis subject's preferred behavior is investment behavior. However, the predetermined task and the preferred behavior are not limited to this example.

[0170] For example, the predetermined task may be increasing sales of a product, and the preferred behavior may be selecting a marketing medium and allocating a budget to each medium. Alternatively, the predetermined task may be acquiring a contract, and the preferred behavior may be allocating time to various sales activities (e.g., visits, phone calls, emails, presentations, etc.).

[0171] Even in the above case, the above-described configuration makes it possible to accurately predict the actions that the subject may perform in a given situation.

[0172] Furthermore, the steps of the processes described in this specification do not necessarily have to be processed in chronological order according to the order shown in the flowcharts or sequence diagrams. For example, the steps of the processes of each device may be processed in an order different from the order shown, or may be processed in parallel.

[0173] Furthermore, the series of processes performed by each device described in this specification may be realized using software, hardware, or a combination of software and hardware. The programs constituting the software may be provided, for example, inside or outside each device and stored in advance in a non-transitory computer-readable medium. Each program is then loaded into RAM when executed by a computer and executed by various processors. Examples of the storage medium include a magnetic disk, an optical disk, a magneto-optical disk, and a flash memory. Furthermore, the computer programs may be distributed, for example, via a network, without using a storage medium.

[0174] Furthermore, the effects described herein are merely descriptive or exemplary and are not limiting. In other words, the technology according to the present disclosure may achieve other effects that are apparent to those skilled in the art from the description of this specification, in addition to or in place of the above-described effects.

[0175] The following configurations also fall within the technical scope of the present disclosure. (1) a prediction unit that outputs preference prediction data that indicates a prediction of a preference behavior that may be performed by the subject in a predetermined situation, based on preference performance data that indicates a performance of the preference behavior related to a predetermined task performed by the subject; Equipped with the prediction unit inputs the preference achievement data into a classifier generated by manifold learning, and outputs the preference prediction data based on applying a prediction model based on assumed information related to the predetermined situation to each of the classified units. Information processing device. (2) the classifier generated by the manifold learning includes a self-organizing map; The information processing device according to (1) above. (3) The preference performance data includes situation transition data indicating a transition of a past situation, and past preference ratio data indicating a preference ratio of preference objects that could have been the target of the preference behavior in the past situation, the prediction unit classifies the preference objects into a plurality of units by inputting the preference record data into the self-organizing map, and obtains a first codebook vector for each unit; The information processing device according to (2) above. (4) the prediction unit applies a prediction model based on assumed information related to the predetermined situation to the first codebook vector acquired for each unit, thereby acquiring a second codebook vector for each unit. The information processing device according to (3) above. (5) the prediction unit performs an inverse transformation process on the second codebook vector acquired for each unit, and outputs the preference prediction data. The information processing device according to (4) above. (6) the preference prediction data includes predicted preference ratio data indicating a predicted preference ratio of the preference object in the predetermined situation; The information processing device according to (5) above. (7) the prediction unit generates a map representing the intensity of the predicted preference ratio of the preferred object for each unit in the form of a heat map based on the predicted preference ratio data; The information processing device according to (6) above. (8) the prediction unit generates a map in which the magnitude of the difference between the predicted preference ratio data and the past preference ratio data is expressed in the form of a heat map for each of the units; The information processing device according to (6) or (7). (9) the prediction unit outputs preference prediction data indicating a prediction of the preference behaviors that may be performed by the plurality of subjects in a predetermined situation, based on the plurality of preference actual data related to the plurality of subjects; The information processing device according to any one of (1) to (8). (10) the predetermined task includes asset management; The preference behavior includes investment behavior in financial products. The information processing device according to any one of (1) to (9). (11) The preference performance data includes performance information of active weights in the past, The preference prediction data includes prediction information of active weights in the predetermined situation. The information processing device according to any one of (1) to (10) above. (12) the expected information includes at least one of expected information on factor returns or expected information on factor properties in the predetermined situation; The information processing device according to any one of (10) and (11). (13) The expected information on the factor returns includes expected information on at least one of a market return, a return difference between value and growth, a return difference between small and large, or momentum. The information processing device according to (12) above. (14) The expected information of the factor properties includes expected information regarding at least one of excess return over a benchmark, market capitalization, or price-to-book ratio. The information processing device according to (12) or (13). (15) The prediction model includes any one of a multiple regression model, a vector autoregression model, or a GNN model. The information processing device according to any one of (1) to (14). (16) a display unit that displays the map generated by the prediction unit; Further provided with The information processing device according to (6) or (7). (17) the processor outputs preference prediction data indicating a prediction of a preference behavior likely to be performed by the subject in a predetermined situation based on preference performance data indicating performance of the preference behavior related to a predetermined task performed by the subject; Including, The outputting includes inputting the preference performance data into a classifier generated by manifold learning, and applying a prediction model based on assumed information related to the predetermined situation to each of the classified units, and outputting the preference prediction data; further comprising: Information processing methods. (18) Computer, a prediction unit that outputs preference prediction data that indicates a prediction of a preference behavior that may be performed by the subject in a predetermined situation, based on preference performance data that indicates a performance of the preference behavior related to a predetermined task performed by the subject; Equipped with the prediction unit inputs the preference achievement data into a classifier generated by manifold learning, and outputs the preference prediction data based on applying a prediction model based on assumed information related to the predetermined situation to each of the classified units. information processing device, A program to function as a [Explanation of symbols]

[0176] 10 Learning Device 110 Learning Department 120 Storage section 20 Prediction Device 210 Prediction Department 220 Storage section 230 Display section 240 Operation section

Claims

1. a prediction unit that outputs preference prediction data that indicates a prediction of a preference behavior that may be performed by the subject in a predetermined situation, based on preference performance data that indicates a performance of the preference behavior related to a predetermined task performed by the subject; Equipped with the prediction unit inputs the preference actual data into a classifier generated by manifold learning that learns classification related to the preference actual data based on the preference actual data, thereby classifying preference objects that could be targets of the preference behavior into a plurality of units and acquiring a first codebook vector for each unit, applies a prediction model based on assumed information related to the predetermined situation to the first codebook vector acquired for each unit, thereby acquiring a second codebook vector for each unit, and outputs the preference prediction data based on the second codebook vector acquired for each unit. Information processing device.

2. the classifier generated by the manifold learning includes a self-organizing map; The information processing device according to claim 1 .

3. The preference record data includes situation transition data indicating a transition of a past situation, and past preference ratio data indicating a preference ratio of the preference object in the past situation. The information processing device according to claim 2 .

4. the prediction unit outputs the preference prediction data by performing an inverse transformation process on the second codebook vector acquired for each unit. The information processing device according to claim 3 .

5. the preference prediction data includes predicted preference ratio data indicating a predicted preference ratio of the preference object in the predetermined situation; The information processing device according to claim 4 .

6. the prediction unit generates a map representing the intensity of the predicted preference ratio of the preferred object for each unit in the form of a heat map based on the predicted preference ratio data; The information processing device according to claim 5 .

7. the prediction unit generates a map in which the magnitude of the difference between the predicted preference ratio data and the past preference ratio data is expressed in the form of a heat map for each of the units; The information processing device according to claim 5 .

8. the prediction unit outputs preference prediction data indicating a prediction of the preference behaviors that may be performed by the plurality of subjects in a predetermined situation, based on the plurality of preference actual data related to the plurality of subjects; The information processing device according to claim 1 .

9. the predetermined task includes asset management; The preference behavior includes investment behavior in financial products. The information processing device according to claim 1 .

10. The preference performance data includes performance information of active weights in the past, The preference prediction data includes prediction information of active weights in the predetermined situation. The information processing device according to claim 1 .

11. the expected information includes at least one of expected information on factor returns or expected information on factor properties in the predetermined situation; The information processing device according to claim 9 .

12. The expected information on the factor returns includes expected information on at least one of a market return, a return difference between value and growth, a return difference between small and large, or momentum. The information processing device according to claim 11.

13. The expected information of the factor properties includes expected information regarding at least one of excess return over a benchmark, market capitalization, or price-to-book ratio. The information processing device according to claim 11.

14. The prediction model includes any one of a multiple regression model, a vector autoregression model, or a GNN model. The information processing device according to claim 1 .

15. a display unit that displays the map generated by the prediction unit; Further provided with The information processing device according to claim 6 .

16. the processor outputs preference prediction data indicating a prediction of a preference behavior likely to be performed by the subject in a predetermined situation based on preference performance data indicating performance of the preference behavior related to a predetermined task performed by the subject; Including, the outputting includes inputting the preference performance data into a classifier generated by manifold learning that learns classification related to the preference performance data based on the preference performance data, thereby classifying preference objects that could be targets of the preference behavior into a plurality of units and acquiring a first codebook vector for each unit, applying a prediction model based on assumed information related to the predetermined situation to the first codebook vector acquired for each unit, thereby acquiring a second codebook vector for each unit, and outputting the preference prediction data based on the second codebook vector acquired for each unit; further comprising: Information processing methods.

17. Computer, a prediction unit that outputs preference prediction data that indicates a prediction of a preference behavior that may be performed by the subject in a predetermined situation, based on preference performance data that indicates a performance of the preference behavior related to a predetermined task performed by the subject; Equipped with the prediction unit inputs the preference actual data into a classifier generated by manifold learning that learns classification related to the preference actual data based on the preference actual data, thereby classifying preference objects that could be targets of the preference behavior into a plurality of units and acquiring a first codebook vector for each unit, applies a prediction model based on assumed information related to the predetermined situation to the first codebook vector acquired for each unit, thereby acquiring a second codebook vector for each unit, and outputs the preference prediction data based on the second codebook vector acquired for each unit. information processing device, A program to function as a

Images