Dwelling environment evaluation system and dwelling environment evaluation method

Abstract

To achieve at high resolution, evaluation of a dwelling environment where taste of an individual is reflected, and provide the evaluation visually.SOLUTION: A dwelling environment evaluation system comprises: an estimation model generation unit configured to cause a model to machine-learn using objective maps generated by a map generation unit as input data, comprehensive happiness degrees and individual environmental factor satisfaction degrees as correct answer data, and priority factors as features, so as to generate an estimation model of the comprehensive happiness degree with priority factors free from bias and an estimation model of the individual environmental factor satisfaction degree; a map generation unit configured to estimate a comprehensive happiness degree map related to comprehensive happiness degrees of inhabitants in a second district using the estimation model of the comprehensive happiness degree, estimate an individual environmental factor satisfaction degree map related to individual environmental factors of inhabitants in the second district using the estimation model of the individual environment factor satisfaction degree, and generating an individualized happiness degree map specialized to inhabitants in the second district using the comprehensive happiness degree map and the individual environment factor satisfaction degree map; and a visualizing unit configured to visualize the individualized happiness degree map.SELECTED DRAWING: None

Description

[Technical field]

[0001] The present invention relates to a housing environment evaluation system and a housing environment evaluation method, and is suitably applied to, for example, a housing environment evaluation system that quantitatively evaluates a housing environment. [Background technology]

[0002] In recent years, there has been a demand for technology to quantitatively evaluate living environments, and research and development into methods for evaluating living environments is underway. Conventional methods for evaluating living environments, for example, refer to quantitative evaluations of living environments based on personal values. This allows real estate agents to shorten the time it takes to conclude a rental or purchase contract, and makes it easier for customers searching for a home (hereinafter referred to as "real estate users") to narrow down the areas that match their desired living environment. Meanwhile, local governments can grasp the current state of an area from a quantitative evaluation of the living environment based on a specified perspective, and use this information in planning facilities, etc.

[0003] Patent Document 1 discloses a living environment evaluation means that outputs real estate information based on weighting of personal values. Patent Document 2 discloses a living environment evaluation device that evaluates living environments based on personal values. The living environment evaluation device disclosed in Patent Document 2 is centered on a model that analyzes open data and calculates indicators related to living environments. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2017-37548 A [Patent Document 2] JP 2017-91422 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, the residential environment evaluation methods disclosed in Patent Document 1 and Patent Document 2 basically do not reflect the happiness level based on the subjective impressions of residents, but evaluate the residential environment based on information at the local level. Evaluation criteria such as average annual income and education level are used as local level information, but such evaluation criteria are not evaluated for the extent to which they actually affect the subjective happiness of residents. On the other hand, when evaluating a residential environment by reflecting the subjective happiness of residents, a survey on the happiness level of the residence is usually conducted directly on the residents. However, it should be noted that the happiness level evaluated during the survey is always biased depending on the local residents because the individual environmental factors that are emphasized are different. In addition, the evaluation of the residential environment needs to be performed at a finer level of granularity than at the local level. This is because if the resolution of the evaluation of the residential environment according to the local level is low, it may take a long time for real estate users to identify a place where they actually want to live.

[0006] The present invention has been made in consideration of the above points, and aims to propose a housing environment evaluation system and a housing environment evaluation method that are capable of visually providing an evaluation of a housing environment that better reflects individual preferences, with higher resolution than conventional methods. [Means for solving the problem]

[0007] In order to solve such problems, the present invention includes a survey information acquisition unit that acquires survey data including an overall happiness level, individual environmental factor satisfaction levels representing satisfaction with individual environmental factors, and important factors, regarding the residential environment in a first region; a map creation unit that uses remote sensing data that can be acquired regarding the residential environment in the first region to generate an objective map representing the individual environmental factor satisfaction levels, and generates a subjective map representing the overall happiness level and the individual environmental factor satisfaction levels; and a map creation unit that uses the objective map as input data, the overall happiness level and the individual environmental factor satisfaction levels as correct answer data, and the important factors as features, to machine-learn a model, and the estimation model generation unit generating an unbiased estimation model of the overall happiness and an estimation model of the individual environmental factor satisfaction; a map generation unit estimating an overall happiness map related to the overall happiness of residents in the second area using the estimation model of the overall happiness and estimating an individual environmental factor satisfaction map related to the individual environmental factors of residents in the second area using the estimation model of the individual environmental factor satisfaction, and generating a personalized happiness map specialized for residents in the second area using the overall happiness map and the individual environmental factor satisfaction map; and a visualization unit visualizing the personalized happiness map.

[0008] Further, in the present invention, a residential environment evaluation method in a residential environment evaluation system that quantitatively evaluates a residential environment includes a step in which a survey information acquisition unit acquires survey data regarding the residential environment in a first region, the survey data including an overall happiness degree, an individual environmental factor satisfaction degree representing the satisfaction of an individual environmental factor, and an important factor; a map creation unit uses remote sensing data that can be acquired regarding the residential environment in the first region to generate an objective map representing the individual environmental factor satisfaction degree and a subjective map representing the overall happiness degree and the individual environmental factor satisfaction degree; and a estimation model generation unit uses the objective map as input data, the overall happiness degree and the individual environmental factor satisfaction degree as correct answer data, and generates the important factor as a feature quantity the step of machine learning a model to generate an estimation model of the overall happiness and an estimation model of the individual environmental factor satisfaction that are unbiased in the priority factors; the step of a map generation unit estimating an overall happiness map related to the overall happiness of residents in the second area using the estimation model of the overall happiness, and estimating an individual environmental factor satisfaction map related to the individual environmental factors of residents in the second area using the estimation model of the individual environmental factor satisfaction, and generating a personalized happiness map specialized for residents in the second area using the overall happiness map and the individual environmental factor satisfaction map; and the step of a visualization unit visualizing the personalized happiness map. Effect of the Invention

[0009] According to the present invention, it is possible to realize a living environment evaluation system and a living environment evaluation method that can visually provide an evaluation of a living environment that better reflects individual preferences with higher resolution than conventional methods. [Brief description of the drawings]

[0010] [Figure 1] 1 is a schematic diagram showing a configuration example of a living environment evaluation system in this embodiment. [Diagram 2] 1 is a configuration diagram showing a hardware configuration of a living environment evaluation system in this embodiment. [Diagram 3]FIG. 2 is a schematic diagram showing the contents of a questionnaire-based survey in this embodiment. [Figure 4] FIG. 2 is a schematic diagram illustrating a database according to the present embodiment. [Diagram 5] 4 is a flowchart illustrating the processing of the entire system in the present embodiment. [Figure 6] 1 is a schematic diagram showing a comprehensive happiness level map and an individual environmental factor satisfaction level map in this embodiment. FIG. [Figure 7] FIG. 2 is a schematic diagram showing a personalized happiness map in the present embodiment. [Figure 8] 11 is a flowchart illustrating a process of a survey information acquisition unit in the present embodiment. [Figure 9] 5 is a flowchart illustrating a process of a remote sensing data analysis unit in the present embodiment. [Figure 10] 11 is a flowchart illustrating a process of a survey information analysis unit in the present embodiment. [Figure 11] 10 is a flowchart illustrating a process of a subjective data generating unit in the present embodiment. [Figure 12] 10 is a flowchart illustrating a process of an objective data generating unit in the present embodiment. [Figure 13] 11 is a flowchart illustrating a process of a learning data generating unit in the present embodiment. [Figure 14] 11 is a flowchart illustrating the processing of a comprehensive happiness level estimation model generation unit in this embodiment. [Figure 15] 5 is a schematic diagram illustrating the processing of an individual environmental factor satisfaction degree estimation model generation unit in the present embodiment. FIG. [Figure 16] FIG. 2 is a schematic diagram showing the processing of a personalized happiness level estimation unit in this embodiment. [Figure 17] 4 is a schematic diagram illustrating the processing of a user-oriented analysis unit in the present embodiment. FIG. [Figure 18] FIG. 13 is a schematic diagram showing a data structure of the overall happiness level learning phase in this embodiment. [Figure 19]FIG. 13 is a schematic diagram showing a data structure of an individual satisfaction level learning phase in the present embodiment. [Figure 20] FIG. 13 is a schematic diagram showing a data structure of the overall happiness level inference phase in this embodiment. [Figure 21] FIG. 2 is a schematic diagram showing a data structure of an individual satisfaction degree inference phase in this embodiment. [Figure 22] 5A to 5C are schematic diagrams illustrating examples of user operations in the present embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] An embodiment of the present invention will now be described in detail with reference to the drawings.

[0012] (1) This embodiment Fig. 1 is a block diagram showing an example of the hardware configuration of a residential environment evaluation system 100 according to this embodiment. The residential environment evaluation system 100 has a configuration in which a remote sensing observation device 120 and a residential environment evaluation device 121 are connected via a network (not shown). The remote sensing observation device 120 is disposed, for example, in a first area where the residential environment is observed. In this embodiment, the geographical area in which the happiness level of residents is estimated is expressed as a "second area" as described later.

[0013] The remote sensing observation device 120 acquires data 101a that can be acquired about the living environment in the first region (hereinafter referred to as "remote sensing data") The remote sensing observation device 120 acquires remote sensing data 101a about images acquired about the surrounding environment of the first region as an example of information about the living environment in the first region, and provides the remote sensing data 101a to the living environment evaluation device 121.

[0014] The remote sensing data 101a is, for example, satellite image data of the earth's surface taken from an observation satellite. The remote sensing data 101a may be of a number of different types, such as aerial photograph data taken from an aircraft, such as RapidEye, MODIS, LandSat, Sentinel, and WorldView, and data related to hearing. The remote sensing data 101a may be, for example, a low-resolution remote sensing image and a high-resolution remote sensing image, but the type of data is not limited to these. The remote sensing data 101a may be time-series images of the same region. In the following embodiment, the remote sensing data 101a will be described by taking satellite image data of the earth's surface taken from an observation satellite as an example.

[0015] The living environment evaluation device 121 is a so-called computer, and is equipped with a CPU, a RAM, a storage unit, etc. In the living environment evaluation device 121, programs and the like stored in the storage unit perform arithmetic processing using the RAM and the like under the control of the CPU.

[0016] Fig. 2 is a block diagram showing an example of the software configuration of the living environment evaluation device 121 shown in Fig. 1. In Fig. 2, the directions of data, signals, and commands between each block are shown by arrows, but this is only an example, and in reality, data, signals, and commands may be sent in a direction different from that of the arrows, or data, signals, and commands may be exchanged between blocks without arrows.

[0017] The living environment evaluation device 121 includes remote sensing data 101a, environmental information data 101b, geographic information data 101c, and survey data 101d as a plurality of different data 101. The living environment evaluation device 121 further includes a remote sensing data acquisition unit 102, a geographic information acquisition unit 103, a survey information acquisition unit 104, a database generation unit 105, a survey information analysis unit 106, a remote sensing data analysis unit 107, an objective data generation unit 108, a subjective data generation unit 109, a learning data generation unit 110, a comprehensive index calculation unit 111, an individual index calculation unit 112, a personalized index calculation unit 113, and a visualization unit 114, and further includes a user-oriented analysis unit 115.

[0018] The remote sensing data acquisition unit 102 acquires remote sensing data 101a and environmental information data 101b, and stores at least the remote sensing data 101a in a database 116. The environmental information data 101b mainly includes vegetation data, surface temperature data, etc., but may also include other data. This vegetation data includes vegetation distribution information, which is information such as latitude and longitude, distribution shape position, etc., but may also be other data.

[0019] The geographic information acquisition unit 103 acquires the geographic information data 101c and stores it in the database 116. The survey information acquisition unit 104 acquires survey data 101d indicating the results of a questionnaire-based survey (see FIG. 3 described later) on the living environment conducted on residents in the first region. The survey data 101d includes an overall happiness level, individual environmental factor satisfaction indicating the satisfaction of individual environmental factors, and an important factor, regarding the living environment in the first region. Specifically, the survey data 101d is question data including evaluation data indicating whether or not the living environment is overall satisfied, environmental factor data regarding the individual environmental factor most important when evaluating the overall living environment, and evaluation data regarding the satisfaction of each individual environmental factor such as the natural environment and shopping convenience, but the contents of the data are not limited thereto.

[0020] The environmental information data 101b is information on the environment obtained from the remote sensing data 101a, for example, when the remote sensing data 101a includes vegetation information, water quality information, soil information, surface temperature information, and the like. This information is vegetation distribution information, spectrum data, water information, soil information, topographical information, and the like. The types of information are not limited to these. The vegetation distribution information is information such as latitude and longitude, distribution shape position, and the like. Among these information, information that can be optically obtained is obtained using an optical sensor. Among optical sensors, particularly passive sensors that can obtain information from light from an object to be observed, basically have the same structure and function as the eye (naked eye), similar to a camera. That is, passive sensors have a function in which an optical system such as a lens (crystalline lens) collects light from an object, forms an image on a detection system (retina), and obtains spectral and spatial information (color and shape).

[0021] In particular, with regard to the spectrum (color), while the human eye can only capture visible light, optical sensors can detect light over a wide range of wavelengths from visible light to infrared light. This makes it possible to obtain a great deal of useful information that cannot be distinguished by the human eye, such as the identification of minerals, rocks, and vegetation, the temperature of the earth's surface, land use, and the water and plankton resources of the sea and lakes. Moreover, it is also possible to obtain this information as two-dimensional images over a wide range. Water information can be extracted from the remote sensing data 101a. For example, remote sensing data such as water surface reflectance and transparency can be analyzed to extract water quality information.

[0022] Examples of water quality information include water temperature, pigment concentration, organic matter content, suspended solids, bottom sediment conditions, etc. Examples of environmental information data 101b include soil data, meteorological data, elevation data, and slope data. Meteorological data includes AMeDAS, MODIS surface temperature satellite, meteorological station data, and the like, but the types are not limited. Geographical information data 101c includes polygons, position information data, and other data that are essential when sharing local data and satellite data.

[0023] The survey data 101d is data collected by a questionnaire-type survey of residents in the first area that is the observation target of the remote sensing data 101a. The questionnaire-type survey includes questions including an evaluation of the overall residential comfort, an evaluation of the degree of importance of individual environmental factors, and an evaluation of each individual environmental factor. Specifically, as shown in FIG. 3, this survey includes a general question, an important question, a first individual question, and a second individual question. The general question is about the satisfaction of the overall residential environment (hereinafter referred to as the "overall residential environment") in the first area, and the important question is about the factor that is most important when evaluating the overall residential environment. The first individual question is about the satisfaction of the natural environment in the first area, and the first individual question is about the satisfaction of the shopping convenience in the first area. Residents of the first area input answers to these questions. The data thus input corresponds to the survey data 101d. Among these, the first individual question and the second individual question correspond to the first individual environmental factor and the second individual environmental factor, respectively, which are included in the important factors described later. The questions are not limited to those listed above and may include other questions.

[0024] The remote sensing data acquisition unit 102 acquires remote sensing data 101a and environmental information data 101b. The geographic information acquisition unit 103 acquires geographic information data 101c. The survey information acquisition unit 104 acquires survey data 101d related to the results of a questionnaire survey of residents in a first region.

[0025] The database generation unit 105 acquires the remote sensing data 101a and the environmental information data 101b from the remote sensing data acquisition unit 102, and stores them in a database 116. The database generation unit 105 stores the geographic information data 101c acquired from the geographic information acquisition unit 103 in the database 116. The database generation unit 105 stores the survey data 101d acquired from the survey information acquisition unit 104 in the database 116.

[0026] The survey information analysis unit 106 analyzes the survey data 101d stored in the database 116 to generate overall happiness survey data. The remote sensing data analysis unit 107 analyzes the remote sensing data 101a stored in the database 116.

[0027] The objective data generation unit 108 is an example of a map creation unit, and uses the remote sensing data 101a to create an objective map representing individual environmental factor satisfaction, which is the satisfaction of the residents of the first region with individual environmental factors such as the "natural environment", and stores the objective map in the database 116. The subjective data generation unit 109 is an example of a map creation unit, and uses the survey data 101d to create a subjective map representing an overall happiness level, which is the overall satisfaction with the living environment of the residents of the first region, and individual environmental factor satisfaction, and stores the subjective map in the database 116. The subjective data generation unit 109 maps the overall happiness level map and the individual environmental factor satisfaction level map (see FIG. 6).

[0028] The learning data generation unit 110 generates learning data using at least one of the objective map created by the objective data generation unit 108 and the subjective map created by the subjective data generation unit 109. The learning data generation unit 110 generates input data and correct answer data as the learning data.

[0029] The learning data generating unit 110 uses, for example, at least a specific index image (objective map) calculated from the remote sensing data 101a related to each environmental factor generated by the objective data generating unit 108, an image for each spectrum based on the remote sensing data 101a, and (if necessary, other related necessary images) as learning data for machine learning to generate a happiness estimation model. In addition, the learning data generating unit 110 uses the overall happiness map and the individual environmental factor satisfaction map to generate correct answer data for learning the happiness estimation model.

[0030] The overall index calculation unit 111 is an example of an estimation model generation unit, which uses an objective map as input data, the overall happiness level and individual environmental factor satisfaction levels as correct answer data, and the important factors as feature quantities to machine-learn a model and generate an estimation model of overall happiness level (hereinafter also referred to as the "overall index calculation model") that is unbiased in the important factors.

[0031] The individual index calculation unit 112 is an example of an estimation model generation unit, which uses the objective map as input data, the overall happiness level and individual environmental factor satisfaction level as correct answer data, and the important factors as features to machine-learn a model and generate an estimation model of individual environmental factor satisfaction level (hereinafter also referred to as the "individual index calculation model") that is unbiased in the important factors.

[0032] The overall index calculation unit 111 and the individual index calculation unit 112 use the objective map 48, which can quantitatively evaluate individual environmental factors, as input data, the overall happiness map as answer data, and the important factors as features, and perform machine learning to generate an estimation model of overall happiness that is unbiased in the important factors.

[0033] The personalization index calculation unit 113 is an example of a map generation unit, which estimates an overall happiness map regarding the overall happiness of residents in the second area using an overall happiness estimation model, and estimates an individual environmental factor satisfaction map regarding individual environmental factors of residents in the second area using an individual environmental factor satisfaction estimation model, and generates a personalized happiness map specialized for residents in the second area using the overall happiness map and the individual environmental factor satisfaction map.

[0034] Specifically, the personalization index calculation unit 113 acquires remote sensing data 101a of the second region. The personalization index calculation unit 113 acquires the overall index calculation model generated by the overall index calculation unit 111. The personalization index calculation unit 113 uses the remote sensing data 101a of the second region and the overall index calculation model to generate an overall happiness map of the second region.

[0035] The personalized index calculation unit 113 acquires the individual index calculation model generated by the individual index calculation unit 112. The personalized index calculation unit 113 uses the remote sensing data 101a of the second region and the individual index calculation model to generate an individual environmental factor satisfaction map of the second region.

[0036] The personalization index calculation unit 113 uses the overall happiness map and the individual environmental factor satisfaction degree map to generate a personalized happiness map. The process of the personalization index calculation unit 113 will be described in detail later.

[0037] The visualization unit 114 visualizes for the user the overall happiness map and the individual environmental factor satisfaction map that indicates the satisfaction with the individual environmental factors (see FIG. 6). The visualization unit 114 also visualizes for the user a personalized happiness map (see FIG. 7) that indicates personalized happiness that reflects personal preferences.

[0038] The user analysis unit 115 has the function of accepting and analyzing feedback from users who have come into contact with the overall happiness map and the individual environmental satisfaction map, or the personalized happiness map, examining the validity of the individual index calculation model and the overall index calculation model, and proposing improvements to the individual index calculation model and the overall index calculation model if necessary.

[0039] FIG. 4 shows an example of various data managed in the database 116 shown in FIG. 2. The database 116 stores at least the geographic information data 101c, the subjective data, the objective data, the important factors, the remote sensing data 101a, the subjective map 47, the objective map 48, the environmental information data 101b, and the survey data 101d. The database 116 may also store a total happiness score, a satisfaction score of an individual environmental factor, a total happiness map, an individual environmental factor satisfaction map, a personalized happiness map, a total spreadsheet model, an individual index calculation model, a happiness estimation model, and the like. In the database 116, an "ID" is an identifier for distinguishing a specific region from other regions, and may be, for example, "ID1", "ID2", or "ID3" according to each region.

[0040] The geographic information data 101c includes an ID for each coordinate. As an example, the subjective data includes an individual environmental factor (a first individual environmental factor in the illustrated example) and a comprehensive happiness index for each ID. The subjective data is data that may be influenced by the resident's subjective evaluation of the living environment. On the other hand, as an example, the objective data includes a first objective index and a second objective index for each coordinate. Unlike the subjective data described above, the objective data is data that is less influenced by the resident's subjective evaluation of the living environment. The important factors include a first individual environmental factor (e.g., "natural environment") and a second individual environmental factor (e.g., "shopping convenience") for each ID.

[0041] The remote sensing data 101a is an accumulation of data acquired by the remote sensing observation device 120 as described above. The subjective map 47 is an accumulation of data generated by the subjective data generating unit 109. The objective map 48 is an accumulation of data generated by the objective data generating unit 108.

[0042] 5 is a flowchart showing an example of the procedure of the living environment evaluation process S301 in this embodiment. This living environment evaluation process S301 corresponds to a living environment evaluation method executed in the living environment evaluation system 100. First, in step S302, the remote sensing data acquisition unit 102 acquires remote sensing data 101a and further acquires environmental information data 101b, the geographic information acquisition unit 103 acquires geographic information data 101c, and the survey information acquisition unit 104 acquires survey data 101d.

[0043] In step S303, the database generation unit 105 acquires the remote sensing data 101a, the environmental information data 101b, the geographic information data 101c, and the survey data 101d, and stores them in the database 116. Here, the geographic information data 101c includes position information and shape information, and for example, masking of a place of interest is performed on the remote sensing data 101a based on the position and shape. Furthermore, the geographic information data 101c may be data on other facilities, such as markets, in addition to data on facilities such as stations that affect the convenience of living for evaluating the living environment.

[0044] The geographic information data 101c includes a masking part. The masking part is extracted, and remote sensing data 101a and environmental information data 101b of the masking part are generated. The database generation unit 105 stores the remote sensing data 101a and environmental information data 101b of the masking part in the database 116.

[0045] In step S304, the remote sensing data analysis unit 107 extracts important factors that are emphasized by residents corresponding to the survey data 101d from the remote sensing data 101a, the environmental information data 101b, and the geographic information data 101c stored in the database 116, based on the survey data 101d stored in the database 116. In this embodiment, a plurality of types of individual environmental factors are exemplified as the important factors. The remote sensing data analysis unit 107 acquires the types of remote sensing data 101a associated with the plurality of individual environmental factors according to the extracted plurality of individual environmental factors, and stores them in the database 116 in association with each other.

[0046] In step S305, the survey information analysis unit 106 analyzes the survey data 101d, first obtaining subjective evaluations of all individual environmental factors from the residents, and then obtaining the residents' subjective evaluation of the overall happiness level under the overall influence of each of the individual environmental factors. Furthermore, the survey information analysis unit 106 obtains the important factors that the residents place the most importance on in evaluating this overall happiness level (individual environmental factors that the residents place particular importance on among the multiple individual environmental factors). In this embodiment, examples of the important factors include "natural environment (first individual environmental factor)" and "shopping convenience (second individual environmental factor)". The remote sensing data analysis unit 107 stores the obtained data on the important factors in the database 116.

[0047] In step S306, the objective data generating unit 108 uses the various individual environmental factors extracted from the survey data 101d stored in the database 116 and the geographic information data 101c to perform calculations from the remote sensing data 101a to obtain various individual environmental factor satisfaction degree maps. Examples of the various individual environmental factor satisfaction degree maps include a normalized vegetation index map representing vegetation distribution information, a surface temperature map representing air temperature, and a heat map representing the distance from a station, but maps based on other indices may also be used. The objective data generating unit 108 stores one or any combination of the various individual environmental factor satisfaction degree maps in the database 116 as an objective map 48. This objective data generating unit 108 uses the remote sensing data 101a to generate the objective map 48 that can quantitatively evaluate the individual environmental factors.

[0048] The objective data generation unit 108 uses an estimation model of overall happiness without bias in priority factors and an objective map 48 that evaluates individual environmental factors from remote sensing data 101a acquired in the second region to estimate an overall happiness map without bias in priority factors for the second region.

[0049] In step S307, the subjective data generating unit 109 maps the satisfaction with various individual environmental factors (individual environmental factor satisfaction), the happiness with respect to the overall residential environment evaluation (overall happiness), the emphasis factors as environmental factors emphasized when evaluating the happiness with respect to the overall residential environment evaluation, and the geographic information data 101c, and also maps the happiness with respect to the overall residential environment evaluation (overall happiness) (see FIG. 6). Furthermore, the subjective data generating unit 109 generates emphasis factor data emphasized when evaluating the happiness with respect to the overall residential environment evaluation (overall happiness).

[0050] The subjective data generation unit 109 generates a satisfaction map for various individual environmental factors (hereinafter referred to as the "individual environmental factor satisfaction map") and a happiness map for the overall living environment evaluation (hereinafter referred to as the "overall happiness map") by the above-mentioned mapping, and stores them in database 116 as subjective map 47, for example, and also stores important factor data in database 116.

[0051] In step S308, the learning data generation unit 110 generates learning data by calculating from the remote sensing data 101a generated in step S306, using the various individual environmental factor satisfaction maps stored in the database 116 and the necessary remote sensing data 101a as input data for machine learning, and using the priority factor data that was given importance when evaluating the happiness level (overall happiness level) for the overall living environment evaluation generated in step S307 as machine learning features.

[0052] The learning data generation unit 110 uses the various individual environmental factor satisfaction maps and the overall happiness map generated in step S307 as correct answer data for machine learning. In this way, the learning data generation unit 110 generates input data and correct answer data as learning data, and provides them to the overall index calculation unit 111 and the individual index calculation unit 112.

[0053] In step S309, the overall index calculation unit 111 performs machine learning on the model using the learning data generated in step S308, and generates an overall index calculation model capable of estimating the happiness level for the overall living environment evaluation. The overall index calculation unit 111 stores the generated overall index calculation model in the database 116. The overall index calculation unit 111 uses the overall happiness level map as correct answer data, the objective map 48 generated from the remote sensing data 101a as input data, and the important factors as feature quantities, performs machine learning on the model, and generates an estimation model of the overall happiness level without bias in the important factors.

[0054] In step S310, the individual index calculation unit 112 performs machine learning on the model using the learning data generated in step S308, and generates an individual index calculation model capable of estimating satisfaction with individual environmental factors. The individual index calculation unit 112 uses the overall happiness map as correct answer data, the objective map 48 generated from the remote sensing data 101a as input data, and the important factors as features, performs machine learning on the model, and generates an individual index calculation model as an estimation model of overall happiness without bias in the important factors. The individual index calculation unit 112 stores the learned individual index calculation model in the database 116.

[0055] In step S311, the personalization index calculation unit 113 uses the overall index calculation model generated in step S309 and the individual index calculation model generated in step S310 to input objective data of various individual environmental factors generated from the remote sensing data 101a obtained in the specified area where the living environment evaluation is desired, and uses the important factor data when evaluating the happiness level for the overall living environment evaluation entered by the user in the second area where the living environment evaluation is desired as features, and generates an overall happiness level map and an individual environmental factor satisfaction level map as shown in FIG. 6.

[0056] The personalization index calculation unit 113 is an example of an estimation model generation unit, and uses an overall happiness map without bias in the priority factors related to the second region and an individual environmental factor satisfaction map estimated using an estimation model of individual environmental factor satisfaction to generate a personalized happiness map (see Figure 7) and store it in the database 116.

[0057] In the overall happiness map and individual environmental factor satisfaction map shown in Figure 6, many regions are indicated by rectangular areas along the X and Y axes and are arranged in a matrix. For example, the region in the upper left is represented by coordinate values ​​of (X1, Y1). For example, the central region in the second row from the top is represented by coordinate values ​​of (X5, Y2). In this overall happiness map and individual environmental factor satisfaction map, the darker the rectangular areas are, the lower the overall happiness and individual satisfaction are, and the lighter the rectangular areas are, the higher the overall happiness and individual satisfaction are.

[0058] On the other hand, in the personalized happiness map shown in Figure 7, for each desired individual environmental factor selected in the upper right corner ("natural environment" and "shopping convenience"), the areas indicated by the darker filled rectangular areas have a lower personalized happiness level, and the areas indicated by the lighter filled rectangular areas have a higher personalized happiness level.

[0059] As an example, the personalization index calculation unit 113 extracts the area with the highest overall happiness level and the area with the highest individual satisfaction level, extracts overlapping areas, and creates a personalized happiness level map.

[0060] Returning to Fig. 5, in step S312, the visualization unit 114 uses a comprehensive happiness map showing the happiness level for the comprehensive living environment evaluation and an individual environment satisfaction map showing the satisfaction level for individual environmental factors to display them on the user interface as shown in Fig. 6. The visualization unit 114 also uses a personalized happiness map showing personalized happiness reflecting personal preferences to display them on the user interface as shown in Fig. 7.

[0061] In step S313, the user-oriented analysis unit 115 receives feedback from users who have come into contact with at least one of the overall happiness map and the individual environmental satisfaction map, examines the validity of the individual index calculation model and the overall index calculation model, and makes a proposal to improve the individual index calculation model and the overall index calculation model as necessary. That is, the user-oriented analysis unit 115 visualizes the overall happiness map without bias of the important factors estimated in the second region, the individual environmental factor satisfaction map, and the personalized happiness map, and presents them to the user, and makes a proposal, for example, to improve the method of acquiring survey information by the survey information acquisition unit 104, or to improve the overall happiness estimation model and the individual environmental factor satisfaction estimation model based on the user's feedback.

[0062] 8 is a flowchart showing an example of the procedure of the survey information acquisition process S401. This survey information acquisition process S401 is executed by the survey information acquisition unit 104. When acquiring survey data 101d, this survey information acquisition process S401 acquires survey data 101d of survey results regarding the happiness of the living environment of residents in a first area. The survey information acquisition unit 104 acquires the survey data 101d including individual environmental factors that were emphasized when evaluating the overall happiness level.

[0063] In the survey method of this embodiment, the survey of happiness level of the living environment is conducted with emphasis on the living environment alone, excluding family situation, income situation, workplace situation, etc. Although these family situation, income situation, workplace situation, etc. are important in determining a person's happiness level, it was noted that it is difficult to reflect their influence in evaluating the living environment.

[0064] The happiness of a living environment is thought to be influenced by various individual environmental factors. The various individual environmental factors include the natural environment, temperature, convenience of living, convenience of educational facilities, and the like, and there are various other individual environmental factors. In this embodiment, satisfaction is investigated from each individual environmental factor, and the individual environmental factors are combined to investigate the overall happiness. Since the survey of residents is biased toward the residents' environmental factors, in order to prevent this bias from affecting the learning of the model, when investigating the overall happiness of the residents, the important factors are investigated as the important individual environmental factors.

[0065] In step S402, the survey information acquisition unit 104 uses the above survey method for the residents to survey the overall happiness score for the living environment. In step S403, the survey information acquisition unit 104 uses the above survey method for the residents to survey the satisfaction score for the individual environmental factors from the viewpoint of the living environment. In step S404, the survey information acquisition unit 104 surveys which individual environmental factors were emphasized when surveying the overall happiness of the living environment for the residents. Note that, regarding the acquisition of the survey data 101d, the survey in the form of a questionnaire is one method, and other methods may be used. In this embodiment, the survey in the form of a questionnaire is taken as an example, and the overall happiness is surveyed first, and the overall happiness is divided into five stages, from unhappiness to happiness, and is scored as the overall question (see FIG. 3). Note that the number of stages and the notation of the happiness are not limited to this. When surveying the overall happiness, the individual environmental factors emphasized in the survey method are surveyed and are set as the important questions (see FIG. 3). Scores are also taken for each individual environmental factor, and are set as the individual questions (first individual question and second individual question) (see FIG. 3). However, the method for setting the questions may be different.

[0066] Next, the database generation unit 105 stores the remote sensing data 101a, the environmental information data 101b, the geographic information data 101c stored in the database 116, and the survey data 101d generated by the survey information acquisition unit 104 together in the database 116. The survey information analysis unit 106 analyzes the generated survey data 101d and provides the database 116 with the results.

[0067] 9 is a flowchart showing an example of the procedure of the remote sensing data analysis process S501. This remote sensing data analysis process S501 is executed by the remote sensing data analysis unit 107. The remote sensing data analysis unit 107 specifies a remote sensing analysis method based on the remote sensing data 101a and the survey data 101d generated by the database generation unit 105.

[0068] The remote sensing data 101a is quantitative data that is objectively observed. In the remote sensing data analysis process S501, the quantitative data is used to estimate the subjective level of happiness, and therefore individual environmental factors related to the subjective level of happiness are represented by the remote sensing data 101a. In this embodiment, the "natural environment" and "shopping convenience" are exemplified as such individual environmental factors. The remote sensing data analysis process S501 identifies objective data that can be generated from the remote sensing data 101a that is related to the individual environmental factors "natural environment" and "shopping convenience."

[0069] In step S502, the remote sensing data analysis unit 107 acquires the remote sensing data 101a generated by the database generation unit 105 and the survey data 101d stored in the database 116. The remote sensing data analysis unit 107 extracts individual environmental factors (e.g., "natural environment" and "shopping convenience") from the survey data 101d.

[0070] In step S503, the remote sensing data analysis unit 107 specifies an analysis method for the remote sensing data 101a from the remote sensing image based on the remote sensing data 101a and the individual environmental factors. Taking the "natural environment" as an example of the individual environmental factor, green vegetation, parks, rivers, etc. are targets for expressing the natural environment. In order to evaluate green vegetation, which is one of the targets, the Normalized Difference Vegetation Index (NDVI) is introduced. The Normalized Difference Vegetation Index (NDVI) is one of the indices used to evaluate the state of vegetation in remote sensing image analysis. The NDVI calculates a value that reflects the physiological state of plants, such as the amount of chlorophyll and photosynthetic activity, by comparing light in the near infrared band with light in the visible band. The NVI is expressed by the calculation formula NVI=(NIR - R) / (NIR + R). NIR represents the reflectance of light in the near infrared band, and R represents the reflectance of light in the red band of visible light. A higher NDVI value indicates a greater amount of vegetation and healthier plants. NDVI is widely used for vegetation management in agriculture and forestry, damage assessment of natural disasters, monitoring of environmental changes, and the like. For example, to evaluate the individual environmental factor "natural environment," the normalized difference vegetation index needs to be generated as one quantitative index from the remote sensing data 101a. Other examples include a heat map of the distance to a park. Note that quantitative indices are not limited to these. The remote sensing data analysis unit 107 identifies the quantitative indices of the individual environmental factors and stores them in the database 116.

[0071] The objective data generation unit 108 uses the remote sensing data 101a stored in the database 116 and the quantitative indicators of the individual environmental factors generated by the survey information analysis unit 106 to generate objective data that serves as input data for generating a happiness estimation model.

[0072] 10 is a flowchart showing an example of the procedure of the survey information analysis process S601. This survey information analysis process S601 is executed by the survey information analysis unit 106. In step S602, the survey information analysis unit 106 calculates an overall happiness score for the resident's living environment based on the survey data 101d from the survey information acquisition unit 104, and stores the overall happiness score in the database 116 in association with the resident's ID.

[0073] In step S603, the survey information analysis unit 106 calculates individual satisfaction scores for individual environmental factors from the perspective of the living environment based on the survey data 101d from the survey information acquisition unit 104, and stores the individual satisfaction scores in the database 116 in association with the resident's ID.

[0074] In step S604, the survey information analysis unit 106 acquires, based on the survey data 101d from the survey information acquisition unit 104, a priority individual index score indicating which individual environmental factor among the important factors was emphasized when investigating the overall happiness of the living environment for the resident, and stores the priority individual index score in the database 116 in association with the resident's ID.

[0075] 11 is a flowchart showing an example of the procedure of objective data generation processing S701. This objective data generation processing S701 is executed by objective data generation unit 108. In step S702, objective data generation unit 108 acquires remote sensing data 101a from database 116. In step S703, objective data generation unit 108 acquires specific objective index data of individual environmental factors as objective data 43 from database 116.

[0076] In step S704, the objective data generating unit 108 uses the remote sensing data 101a and the specific objective index data of the individual environmental factors to generate an objective and quantitative individual environmental index image reflecting the individual environmental factors as the objective map 48. For example, if the individual environmental factor is the "natural environment", the specific index is the normalized difference vegetation index. The objective data generating unit 108 generates a normalized difference vegetation index image from a multispectral satellite image. The spectral information of the multispectral satellite image includes R, G, B, near infrared, etc. If the remote sensing data 101a used is different, the spectral information is also different. The objective data generating unit 108 calculates the normalized difference vegetation index using the R and near infrared of the spectral information. The objective data generating unit 108 calculates the normalized difference vegetation index for each pixel and then generates a normalized difference vegetation index image. Other specific objective index data of the individual environmental factors can be adopted, and the calculation method is not limited to this.

[0077] 12 is a flowchart showing an example of the procedure of the subjective data generation process S801. This subjective data generation process S801 is executed by the subjective data generation unit 109. The subjective data generation unit 109 generates subjective data 45 representing subjective happiness for learning a happiness estimation model, using the overall happiness score stored in the database 116, the satisfaction scores of the individual environmental factors, and the important factor data when evaluating the overall happiness.

[0078] In step S802, the overall happiness survey data generated from the survey data 101d by the survey information analysis unit 106 is input to the subjective data generation unit 109. The overall happiness survey data includes information for each resident ID, and this information includes geographical information.

[0079] In step S803, the subjective data generation unit 109 extracts geographic information from the overall happiness survey data using the information for each resident's ID generated by the survey information analysis unit 106. When investigating happiness, other information may be included in addition to the happiness score, and the resident's place of residence is one of them. To protect personal information, the granularity of the resident's place of residence varies, but the subjective data generation unit 109 extracts geographic information for each ID according to the resident's place of residence.

[0080] As an example, depending on the assumed resolution of the happiness estimation, if it is assumed to be 1 square kilometer, residential areas can be identified at this granularity of 1 square kilometer. If the residential granularity of the happiness survey is assumed to be greater than 1 square kilometer, it is also possible to estimate geographic information by residential area. An example of an estimation method is a method that uses Gaussian PDF to randomly distribute geographic information about residents. Gaussian PDF (Gaussian Probability Density Function) is a probability density function that represents a normal distribution (Gaussian Distribution) in statistics and probability theory. A normal distribution is defined by two parameters, the mean and the standard deviation, and shows a bell-shaped distribution that is symmetrical around the mean value. The normal distribution is known to appear in many natural and social phenomena, and is an important distribution in statistics. The estimation method is not limited to this, and other methods may be used.

[0081] In step S804, the subjective data generation unit 109 combines the extracted overall happiness for each resident with the geographic information data 101c of the area where the resident lives to generate an overall happiness map (see FIG. 6). The subjective data generation unit 109 uses the assumed happiness granularity, and assuming 1 square kilometer as an example, the resolution of the map will be 1 square kilometer. One pixel is 1 square kilometer, and if a resident is assigned to that pixel, the value of this pixel is the happiness of the resident. If multiple residents are concentrated in one pixel, the statistical calculation result is set as the value of the pixel. Possible statistical calculations include, but are not limited to, the average value, median value, etc.

[0082] In step S805, the subjective data generation unit 109 generates an individual environmental factor satisfaction map by combining the extracted satisfaction score for the individual environmental factor "natural environment" for each resident with the resident's geographic information (see FIG. 6). The generation method can refer to the case of the overall happiness map. In step S806, the subjective data generation unit 109 generates important factor data for evaluating the overall happiness generated by the survey information analysis unit 106.

[0083] The learning data generation unit 110 uses the subjective map 47 generated by the subjective data generation unit 109, the important factor data, and the objective map 48 generated by the objective data generation unit 108 to generate learning data for generating a happiness level estimation model.

[0084] 13 is a flowchart showing an example of the procedure of the learning data processing S901. This learning data processing S901 is executed by the learning data generating unit 110. In step S902, the learning data generating unit 110 acquires input data. Specifically, the learning data generating unit 110 uses at least a specific index image (objective map 48) calculated from the remote sensing data 101a related to each individual environmental factor generated by the objective data generating unit 108, an image for each spectrum based on the remote sensing data 101a, and (if necessary, other related images) as learning data for learning a model for estimating happiness.

[0085] The generation of input data differs depending on the machine learning model. As an example, when the machine learning model is a CNN (Convolutional Neural Network) described later, all images are treated as bands and synthesized into one multiband image. The method of generating input data is not limited to this. In addition to the multiband image of the input data, the important factor data 44 generated by the subjective data generating unit 109 is used to generate a feature quantity representing the important factor. The learning data generating unit 110 uses the feature quantity representing the important factor and the multiband image as input data for the learning data.

[0086] In step S903, the learning data generating unit 110 acquires correct answer data. Specifically, the learning data generating unit 110 generates correct answer data for learning the happiness estimation model using the overall happiness map and the individual environmental factor satisfaction map generated by the subjective data generating unit 109. When generating the happiness map, the values ​​of the pixels constituting the happiness map are regarded as the scores of the residents who were surveyed, but there are cases where all pixels in the happiness map cannot be filled. Therefore, in order to generate a more reasonable happiness map, the happiness scores are filled in for all pixels in the happiness map by interpolating between pixels having scores.

[0087] Pixel value interpolation refers to estimating the value of a missing pixel in a happiness map by using the values ​​of surrounding pixels when the pixel value is missing. If there are known pixels around a pixel with a missing pixel value, an interpolation algorithm can be used to estimate the value of the missing pixel. Interpolation algorithms can improve the appearance and analysis of happiness maps with many missing pixels.

[0088] Methods for interpolating pixel values ​​include nearest neighbor interpolation, linear interpolation, and bicubic interpolation. Nearest neighbor interpolation is a method of interpolating using the value of a known pixel that is closest to the pixel to be interpolated. Nearest neighbor interpolation may produce a stepped interpolation result, which may degrade image quality, but it is easy to calculate and implement. Linear interpolation is a method of performing linear interpolation using the values ​​of four points surrounding the pixel to be interpolated. The value of a point on a line segment connecting the four surrounding points is found, and this value is used as the interpolation result. This linear interpolation method has higher interpolation accuracy than nearest neighbor interpolation, and can obtain an image with a smoother interpolation result. Bicubic interpolation is a method of performing three-dimensional polynomial interpolation using the values ​​of 16 points surrounding the pixel to be interpolated. The expression of a three-dimensional polynomial connecting the 16 surrounding points is found, and the interpolated value is found from this expression. This bicubic interpolation is a method of performing interpolation with higher interpolation accuracy than linear interpolation, and can obtain a smoother, more natural image. However, due to the large amount of calculations, it may take a long time to process. Interpolation methods are not limited to these.

[0089] 14 is a flowchart showing an example of the procedure of the overall index calculation model generation process. This overall index calculation model generation process is executed by the overall index calculation unit 111. The overall index calculation unit 111 uses the input data and correct answer data generated by the learning data generation unit 110 to machine-learn a model and generate an overall happiness model as an estimation model of the overall happiness.

[0090] In step S1002, the overall index calculation unit 111 inputs the input data generated by the learning data generation unit 110. In this step S1002, the overall index calculation unit 111 specifies a model for estimating the overall happiness level. There are various types of machine learning models, but the most representative machine learning model for estimating an image is CNN, also known as a convolutional neural network. CNN (Convolutional Neural Network) is a type of neural network and is one of the deep learning methods that exhibits high performance mainly in image processing. CNN is composed of a convolutional layer, a pooling layer, and a fully connected layer. The convolutional layer performs a convolution operation to capture the features of the image. The pooling layer performs sampling to compress the features from the feature map obtained by the convolution operation. The fully connected layer is used to output the final classification or regression results based on the information obtained from the feature map. CNN is widely used in fields other than image processing, such as speech and natural language processing, and occupies a very important position in the field of deep learning. There are various models other than CNN, and many models have evolved from CNN. Models are not limited to these.

[0091] In step S1003, the overall index calculation unit 111 acquires a model before machine learning. In step S1004, the overall index calculation unit 111 acquires the correct answer data generated by the learning data generation unit 110. In step S1005, the overall index calculation unit 111 performs machine learning on the model acquired as described above using the correct answer data generated by the learning data generation unit 110, the input data generated by the learning data generation unit 110, and the important factor data 44, to generate an overall happiness model.

[0092] 15 is a flowchart showing an example of the procedure of the individual index calculation model generation process S1101. This individual index calculation model generation process S1101 is executed by the individual index calculation unit 112. In step S1102, the individual index calculation unit 112 acquires input data generated by the learning data generation unit 110.

[0093] In step S1102, the individual index calculation unit 112 acquires a model for estimating individual environmental factor satisfaction. In step S1103, the individual index calculation unit 112 acquires correct answer data of the overall happiness level generated by the learning data generation unit 110. In step S1104, the individual index calculation unit 112 generates learning data using the correct answer data, the input data generated by the learning data generation unit 110, and the above specified model. In step S1105, the individual index calculation unit 112 uses this learning data to machine-learn a model and generates an individual index calculation model as an estimated model of individual environmental factor satisfaction.

[0094] 16 is a flowchart showing an example of the procedure of the personalized index generation process S1201. This personalized index generation process S1201 is executed by the personalized index calculation unit 113. In step S1202, the personalized index calculation unit 113 acquires remote sensing data 101a of a second region.

[0095] In step S1204, the personalization index calculation unit 113 acquires the overall index calculation model generated by the overall index calculation unit 111. In step S1206, the personalization index calculation unit 113 uses the remote sensing data 101a of the second region and the overall index calculation model to generate an overall happiness map 47a of the second region (see the center of FIG. 22).

[0096] Meanwhile, in step S1203, the personalization index calculation unit 113 acquires the individual index calculation model generated by the individual index calculation unit 112. In step S1205, the personalization index calculation unit 113 uses a remote sensing image based on the remote sensing data 101a of the second region and the individual index calculation model to generate an individual environmental factor satisfaction map 47b of the second region (see the left side of FIG. 22).

[0097] In step S1207, the personalization index calculation unit 113 uses the overall happiness map and the individual environmental factor satisfaction map to generate the personalized happiness map 47c (see the right side of FIG. 22). In this personalized happiness map 47c, it is considered that residents living in the area 99 where the overall happiness map 47a and the individual environmental factor satisfaction map 47b overlap tend to feel that their living environment is good.

[0098] FIG. 17 is a flowchart showing an example of the procedure of the user-oriented analysis process S1301. This user-oriented analysis process S1301 is executed by the user-oriented analysis unit 115. In step S1302, the user-oriented analysis unit 115 presents a personalized happiness map 47c based on the user's important factors (for example, the individual environmental factor of "natural environment") using various maps (individual environmental factor satisfaction map 47b, overall happiness map 47a, personalized happiness map 47c) estimated as shown in FIG. 22. In step S1303, the user-oriented analysis unit 115 uses the presented personalized happiness map 47c to receive feedback from the user about the state of the living environment.

[0099] In step S1303, the user-oriented analysis unit 115 reflects this feedback and proposes improvements to the survey information acquisition method, such as improvements and additions of individual environmental factors. In step S1304, the user-oriented analysis unit 115 improves the model performance of the comprehensive index calculation model and the individual index calculation model through improvements to the survey information acquisition method.

[0100] 18 is a schematic diagram showing an example of a data structure of the learning phase of the comprehensive index calculation model. The objective data generating unit 108 generates objective data 43 from the remote sensing data 101a, and takes as input data an objective map 48 of a specific index related to an individual environmental factor. The objective data generating unit 108 also takes as features the important factor data 44 obtained from the survey data 101d.

[0101] The objective data generation unit 108 combines the subjective data 45 obtained from the survey data 101d with the geographic information data 101c to generate a subjective map 47, which is treated as correct answer data. The objective data generation unit 108 performs machine learning on a model using the input data, correct answer data, and feature quantities, and generates a comprehensive happiness model as a happiness estimation model 49.

[0102] 19 is a schematic diagram showing the data structure of the learning phase of the individual index calculation model. The objective data generation unit 108 generates objective data 43 from the remote sensing data 101a, and uses the objective map 48 of the specific index related to the individual environmental factor as input data. The subjective data generation unit 109 combines the subjective data 45 obtained from the survey data 101d with the geographic information data 101c to generate the subjective map 47, which is used as the correct answer data. The learning data generation unit 110 uses the input data and the correct answer data to machine-learn the model, and generates an individual index calculation model as a happiness degree estimation model 49.

[0103] 20 is a schematic diagram showing an example of a data structure in the overall happiness inference phase. The overall index calculation unit 111 uses the remote sensing data 101a of the second region to generate an objective map 48. The overall index calculation unit 111 uses the important factors with the content that there is no important item for the important factor data 44, inputs the objective map 48 and the important factor data 44 as input data, performs machine learning on the model, and generates a happiness estimation model 49. The overall index calculation unit 111 generates an overall happiness map 47a of the second region.

[0104] 21 is a schematic diagram showing an example of a data structure of the individual satisfaction inference phase. The individual index calculation unit 112 uses the remote sensing data 101a of the second region to generate an objective map 48. The individual index calculation unit 112 uses the objective map 48 as input data to machine-learn a model and generate an individual satisfaction estimation model 16. The individual index calculation unit 112 generates an individual environmental factor satisfaction map 47b of the second region.

[0105] 22 is a schematic diagram showing an example of a GUI (Graphical User Interface) when presented to a user by the visualization unit 114. The visualization unit 114 visualizes an estimated overall happiness map 47a and an individual environmental factor satisfaction degree map 47b, and further visualizes a personalized happiness map 47c.

[0106] First, a user selects a preferred individual environmental factor. In the illustrated example, the individual environmental factor "natural environment" is selected from among the important factors including the individual environmental factors "natural environment" and "shopping convenience." The visualization unit 114 visualizes, through the GUI, an individual environmental factor satisfaction map 47b (the "important factor map" illustrated), a total happiness map 47a, and a personalized happiness map 47c that influences the individual environmental factor selected as described above, according to the selected individual environmental factor. Note that the detailed design and detailed items of the GUI are not limited to the illustrated example, and may be displayed in other manners.

[0107] In the individual environmental factor satisfaction map 47b, a recommended area with a high score is estimated. In the overall happiness map 47a, a recommended area with a high overall happiness score is estimated. In the personalized happiness map 47c, an overlapping area between the recommended area in the individual environmental factor satisfaction map 47b and the recommended area in the overall happiness map 47a is extracted and set as a recommended area 99 in the personalized happiness map 47c. The extraction of the recommended area 99 in the personalized happiness map 47c may be performed by other methods.

[0108] In the illustrated example, the individual environmental factor satisfaction map 47b, the overall happiness map 47a, and the personalized happiness map 47c all indicate, for example, recommended areas for a user searching for a house by filling them in, making it easy for the user to understand which area they would feel would be happier if they lived there.

[0109] The living environment evaluation system according to this embodiment includes a survey information acquisition unit 104 that acquires survey data 101d including an overall happiness level, individual environmental factor satisfaction levels representing satisfaction with individual environmental factors, and important factors, regarding the living environment in a first region; an objective data generation unit 108 that generates an objective map 48 representing the individual environmental factor satisfaction levels, using remote sensing data 101a that can be acquired regarding the living environment in the first region; a subjective data generation unit 109 that generates a subjective map 47 representing the overall happiness level and the individual environmental factor satisfaction levels, using the remote sensing data 101a that can be acquired regarding the living environment in the first region; and a subjective data generation unit 109 that uses the objective map 48 as input data, the overall happiness level and the individual environmental factor satisfaction levels as correct answer data, and generates the important factor data 45. The system further comprises an overall index calculation unit 111 and an individual index calculation unit 112 which use machine learning to train a model using feature vector 4 as a feature parameter to generate an overall happiness estimation model and an individual environmental factor satisfaction estimation model that are unbiased in terms of priority factors, a personalization index calculation unit 113 which uses the overall happiness estimation model to estimate an overall happiness map related to the overall happiness of residents in the second area, and uses the individual environmental factor satisfaction estimation model to estimate an individual environmental factor satisfaction map related to individual environmental factors of the residents in the second area, and generates a personalized happiness map 47c specialized for residents in the second area using the overall happiness map 47a and the individual environmental factor satisfaction map 47b, and a visualization unit 114 which visualizes the personalized happiness map.

[0110] In the residential environment evaluation method of the residential environment evaluation system according to this embodiment, the survey information acquisition unit 104 acquires survey data 101d including an overall happiness level, individual environmental factor satisfaction level representing satisfaction of individual environmental factors, and important factors, regarding the residential environment in a first region; the objective data generation unit 108 generates an objective map 48 representing the individual environmental factor satisfaction level using remote sensing data that can be acquired regarding the residential environment in the first region, and the subjective data generation unit 109 generates a subjective map 47 representing the overall happiness level and the individual environmental factor satisfaction level; an overall index calculation unit 111 as an example of an estimation model generation unit machine-learns a model using the objective map 48 as input data, the overall happiness level and the individual environmental factor satisfaction level as correct answer data, and the important factors as feature quantities, to generate an estimation model of the overall happiness level without bias in the important factors; and an individual index calculation unit 112 as an example of an estimation model generation unit. The system further comprises a step of: using the objective map 48 as input data, the overall happiness level and the individual environmental factor satisfaction level as correct answer data, and the priority factors as feature quantities, machine learning a model to generate an estimated model of individual environmental factor satisfaction that is unbiased in the priority factors; a step of an overall index calculation unit 111 as an example of a map generation unit estimating an overall happiness level map regarding the overall happiness of residents in the second area using the overall happiness level estimation model; an individual index calculation unit 112 as an example of a map generation unit estimating an individual environmental factor satisfaction level map regarding the individual environmental factors of residents in the second area using the individual environmental factor satisfaction estimation model; a personalization index calculation unit 113 as an example of a map generation unit using the overall happiness level map and the individual environmental factor satisfaction level map to generate a personalized happiness level map specialized for residents in the second area; and a step of a visualization unit 114 visualizing the personalized happiness level map.

[0111] In this way, it is possible to visually provide an evaluation of the living environment that better reflects individual preferences based on individual environmental factors, with a higher resolution than ever before.

[0112] In this embodiment, the survey information acquisition unit 104 acquires survey data 101d including the individual environmental factors that were emphasized when evaluating the overall happiness level. In this way, it is possible to realize a high-resolution evaluation of the living environment that better reflects the individual's preferences by using the individual environmental factors included in the survey data 101d.

[0113] In this embodiment, an objective data generating unit 108, which is an example of a map creating unit, generates an objective map 48 that can quantitatively evaluate individual environmental factors by using the remote sensing data 101a. In this way, it is possible to realize an objective evaluation of the living environment with high resolution that better reflects individual preferences.

[0114] In this embodiment, the subjective data generating unit 109, which is an example of a map creating unit, maps the overall happiness level and individual environmental factors. In this way, it is possible to realize a subjective evaluation of the living environment with high resolution that better reflects individual preferences.

[0115] In this embodiment, the overall index calculation unit 111 and the individual index calculation unit 112, which are examples of an estimation model generation unit, use the objective map 48 that can quantitatively evaluate individual environmental factors as input data, the overall happiness map as answer data, and the important factors as feature quantities, and perform machine learning to generate an estimation model of overall happiness that is not biased in the important factors. In this way, since the important factors are used as feature quantities, it is possible to generate an estimation model of overall happiness that is more unbiased in the important factors.

[0116] In this embodiment, the objective data generating unit 108, which is an example of a map generating unit, uses an estimation model of overall happiness without bias in the important factors, and uses an objective map that evaluates individual environmental factors from the remote sensing data 101a acquired in the second region, to estimate an overall happiness map without bias in the important factors for the second region. In this way, an objective evaluation of the living environment that reflects individual preferences can be realized with high resolution.

[0117] In this embodiment, the personalization index calculation unit 113, which is an example of a model generation unit, generates a personalized happiness map using a general happiness map without bias of the important factors related to the second region and an individual environmental factor satisfaction map estimated using an individual environmental factor satisfaction estimation model. In this way, an objective evaluation of the living environment that reflects individual preferences can be realized with high resolution.

[0118] In this embodiment, the overall index calculation unit 111 and the individual index calculation unit 112, which are examples of an estimation model generation unit, use the overall happiness map as correct answer data, the objective map 48 generated from the remote sensing data 101a as input data, and the important factors as feature quantities, and perform machine learning on the model to generate an estimation model of overall happiness without bias in the important factors. In this way, since the important factors are used as feature quantities, it is possible to generate an estimation model of overall happiness without bias in the important factors.

[0119] In this embodiment, the user analysis unit 115 visualizes and presents to the user an overall happiness map without bias of the important factors estimated in the second region, an individual environmental factor satisfaction map, and a personalized happiness map, and proposes to improve the method of acquiring survey information by the survey information acquisition unit and to improve the overall happiness estimation model and the individual environmental factor satisfaction estimation model using the user's feedback. Such feedback makes it possible to realize a high-resolution evaluation of the living environment that better reflects individual preferences.

[0120] (6) Other embodiments In the above embodiment, the remote sensing observation device 120 is exemplified as a sensor, but the present invention is not limited to this, and other types of data may be acquired as long as the data is capable of evaluating the living environment. [Industrial Applicability]

[0121] The present invention can be applied to a living environment evaluation system that quantitatively evaluates a living environment. [Explanation of symbols]

[0122] 100...living environment evaluation system, 101a...remote sensing sensor, 101b...environmental information data, 101c...geographical information data, 101d...survey data, 102...remote sensing acquisition unit, 103...geographical information acquisition unit, 104...survey information acquisition unit, 105...database generation unit, 106...survey information analysis unit, 107...remote sensing data analysis unit, 108...objective data generation unit, 109...subjective data generation unit, 110...learning data generation unit, 111...overall index calculation unit, 112...individual index calculation unit, 113...personalized index calculation unit, 114...visualization unit, 115...user-oriented analysis unit, 116...database.

Claims

1. a survey information acquisition unit that acquires survey data including an overall happiness level, individual environmental factor satisfaction levels indicating satisfaction with individual environmental factors, and important factors, regarding the living environment in the first region; a map creation unit that uses remote sensing data that can be acquired regarding the living environment in the first region to create an objective map that represents the individual environmental factor satisfaction level, and also creates a subjective map that represents the overall happiness level and the individual environmental factor satisfaction level; an estimation model generation unit that uses the objective map as input data, the overall happiness level and the individual environmental factor satisfaction level as correct answer data, and the important factors as feature quantities to machine-learn a model and generate an estimation model of the overall happiness level and an estimation model of the individual environmental factor satisfaction level that are not biased with respect to the important factors; a map generating unit that uses the estimation model of the overall happiness to estimate an overall happiness map regarding the overall happiness of residents in a second area, and uses the estimation model of the individual environmental factor satisfaction to estimate an individual environmental factor satisfaction map regarding the individual environmental factors of the residents in the second area, and generates a personalized happiness map specialized for the residents in the second area using the overall happiness map and the individual environmental factor satisfaction map; A visualization unit that visualizes the personalized happiness map; A living environment evaluation system comprising:

2. The investigation information acquisition unit Obtaining the survey data including the individual environmental factors that were emphasized when evaluating the overall happiness level; 2. The living environment evaluation system according to claim 1 .

3. The map creation unit uses the remote sensing data to generate the objective map capable of quantitatively evaluating the individual environmental factors.

3. The living environment evaluation system according to claim 2, wherein:

4. The map creation unit is Mapping the overall happiness level and the individual environmental factors 2. The living environment evaluation system according to claim 1,

5. The estimation model generation unit The objective map capable of quantitatively evaluating the individual environmental factors is used as input data, the overall happiness map is used as answer data, the important factors are used as features, and machine learning is performed to generate an estimation model of the overall happiness that is not biased in the important factors.

2. The living environment evaluation system according to claim 1,

6. The map generation unit is Using an estimation model of the overall happiness degree that is unbiased in the importance factor, using the objective map to estimate the individual environmental factors from the remote sensing data acquired in the second region; Estimating the overall happiness map without bias of the importance factor for the second region 2. The living environment evaluation system according to claim 1 .

7. The estimation model generation unit The overall happiness map without bias in the importance factor for the second region; Using the individual environmental factor satisfaction map estimated using the estimation model of the individual environmental factor satisfaction, generate the personalized happiness map.

2. The living environment evaluation system according to claim 1 .

8. The estimation model generation unit The overall happiness map is used as correct answer data, the objective map generated from the remote sensing data is used as input data, the important factors are used as features, and a model is machine-learned to generate an estimation model of the overall happiness without bias in the important factors.

6. The living environment evaluation system according to claim 5.

9. The overall happiness map without bias of the importance factor estimated in the second region, the individual environmental factor satisfaction map, and the personalized happiness map are visualized and presented to a user; A user-oriented analysis unit that uses user feedback to improve the method of acquiring the survey data by the survey information acquisition unit and makes suggestions to improve the estimation model of the overall happiness level and the estimation model of the individual environmental factor satisfaction level.

9. The living environment evaluation system according to claim 8.

10. In a method for evaluating longevity environment in a living environment evaluation system for quantitatively evaluating a living environment, A survey information acquisition unit acquires survey data 101d including an overall happiness level, individual environmental factor satisfaction levels indicating satisfaction levels of individual environmental factors, and important factors, related to the living environment in the first region; a map creation unit generating an objective map representing the individual environmental factor satisfaction level and a subjective map representing the overall happiness level and the individual environmental factor satisfaction level using remote sensing data that can be acquired regarding the living environment in the first region; an estimation model generation unit performs machine learning on a model using the objective map as input data, the overall happiness level and the individual environmental factor satisfaction level as correct answer data, and the important factors as feature quantities, to generate an estimation model of the overall happiness level and an estimation model of the individual environmental factor satisfaction level that are not biased in the important factors; a map generation unit estimating an overall happiness map regarding the overall happiness of residents in a second area using the estimation model of the overall happiness, and estimating an individual environmental factor satisfaction map regarding the individual environmental factors of residents in the second area using the estimation model of the individual environmental factor satisfaction, and generating a personalized happiness map specialized for residents in the second area using the overall happiness map and the individual environmental factor satisfaction map; A visualization unit visualizes the personalized happiness map; A living environment evaluation method comprising the steps of:

Images