JP5965267B2 - Information processing apparatus and information processing method - Google Patents

Description

  The present invention relates to information processing technology, and relates to information processing technology for estimating frequency based on frequency distribution of different areas, for example, creating information for supporting analysis activities for estimating migrants' characteristics from regional information It is related to the technology.

The frequency distribution table is widely used academically and practically.
For example, in order to deepen business relationships with customers and increase customer loyalty in corporate sales activities, analysis work that grasps the characteristics of each region for the purpose of determining the uniqueness and heterogeneity of the region from regional information is required. It has been. For this reason, there is a need for an analysis technique that extracts information useful for grasping regional characteristics from demographics, financial information, and consumer purchasing power information that are tabulated by region.

  Conventionally, there is an analysis method aiming at sales activity support using local information. For example, in Patent Document 1 described below, information useful for a merchandise seller is generated by analyzing sales performance information based on information on a region where the purchaser of the merchandise resides.

JP 2009-169699 A

  By the way, in general, when performing analysis based on regional information, it is necessary to subdivide the region. This is to find out the needs of the region by focusing on the subtle changes by segmenting the market into smaller ones. For this reason, when performing analysis based on regional information, it is effective to subdivide into small units such as towns and villages rather than large groups such as prefectures and municipalities.

  However, there is a database in which regional information is collected that contains detailed information only up to the municipality unit, and detailed information is missing in the town-by-town unit.

  For this reason, regional information is analyzed only in the upper-level area units such as municipalities, and changes in the market in the lower-level area units such as town-by-town cannot be found and appropriate judgment and improvement actions cannot be taken. It becomes a state.

  In addition, it is useful if the frequency distribution data can be interpolated between different measurement areas when the frequency obtained by experiments is used as the gray level and the frequency for each gray level is obtained to obtain the frequency distribution.

  An object of the present invention is to perform an information process for interpolating the frequency distribution by obtaining the relationship of the frequency distribution between a large area and a small area different from the large area.

  According to an aspect of the present invention, the first representative value of the first frequency distribution in the first region and the second representative value of the second frequency distribution in the second region different from the first region. In the representative value comparison unit and the representative value comparison unit, the first frequency distribution and the second frequency distribution when the degree of coincidence between the first representative value and the second representative value is low. And a new second representative value based on the second frequency distribution converted by the distribution conversion unit and a distribution conversion unit that converts the second frequency distribution in a direction that reduces the distribution bias The new second representative value is compared with the first representative value by the representative value comparison unit, and the degree of coincidence between the first representative value and the new second representative value is high. There is provided an information processing apparatus comprising: a distribution estimation unit that continues the process of obtaining the second representative value until

  It has a frequency distribution calculation unit that calculates the frequency of the second frequency distribution based on the second representative value and the frequency distribution estimated by the distribution estimation unit.

  In addition, the present invention is an information processing apparatus provided with a regional information database, which compares two average values of a region in a higher section and a region in a lower section that are stored in advance in the region information database. And a conversion unit for obtaining a conversion method for approximating the frequency distribution of the upper section region to the frequency distribution of the lower section region based on the comparison result, and the lower order from the frequency distribution of the upper section region using the conversion method An information processing apparatus comprising: an estimation unit that performs processing for estimating a frequency distribution of a partition; and a bootstrap processing unit that performs bootstrap processing for obtaining an average value of a population from the estimated frequency distribution is there. Thereby, it is possible to accurately estimate the frequency distribution of the area of the lower section.

  According to another aspect of the present invention, the first representative value of the first frequency distribution in the first region and the second representative of the second frequency distribution in the second region different from the first region. In the representative value comparison step for comparing values and the representative value comparison step, when the degree of coincidence between the first representative value and the second representative value is low, the first frequency distribution and the second value distribution A distribution conversion step of converting the second frequency distribution in a direction to reduce the deviation of the distribution from the frequency distribution, and a new second frequency distribution based on the converted second frequency distribution converted by the distribution conversion step. A representative value is obtained, the new second representative value and the first representative value are compared by the representative value comparison unit, and the degree of coincidence between the first representative value and the new second representative value And a distribution estimation step of continuing the process of obtaining the second representative value until the value becomes high. The information processing method according to symptoms is provided.

  The present invention may be a program for causing a computer to execute the information processing method described above, or a computer-readable recording medium for recording the program.

The present invention has the following effects.
Even if missing information parts exist in attribute value information in different areas, the missing information part is estimated based on one frequency distribution, attribute average value, and other attribute average value, and provided to the user. Analysis can be estimated.

1 is a system configuration diagram including an information processing apparatus according to an embodiment of the present invention. It is a figure which shows an example of the frequency distribution used for the information processing by this Embodiment. It is a flowchart figure which shows the flow of the information processing by this Embodiment. It is a flowchart which shows the outline | summary of the information processing performed within a host system. FIG. 4 (a) shows data obtained by summing up deposit / saving amount information in units of municipalities such as A city and B city corresponding to the upper division area. FIG. 4 (b) shows data obtained by tabulating deposit and saving amount information in units of towns, such as C town of A prefecture B city, which is a subdivision area of A prefecture B city. FIG. 5 is a flowchart showing an outline of processing in the host system 13 as an example of estimating the number of populations according to deposit and saving amounts in FIG. It is the image figure which showed the difference of the conversion formula f (x) by the comparison result of E (a) and E (b). It is an image figure of the process which estimates a missing information part from the population distribution h (b) according to the deposit and saving amount of the calculated A prefecture B city C town. It is a figure which shows the example of the frequency distribution used for the information processing by this Embodiment, A target is LED formed in large numbers on a semiconductor substrate, a class is a light emission wavelength band, and frequency corresponds to a semiconductor substrate. It is the number of LEDs having an emission wavelength band.

Hereinafter, an information processing technique according to an embodiment of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a system configuration diagram including an information processing apparatus according to an embodiment of the present invention. The information system according to the present embodiment is a frequency distribution (probability distribution) information database relating to characteristics of regional information and the like (how much frequency a certain characteristic value exists), here as an example, as described later in FIG. A host system 13 such as a personal computer provided with a database 11 is provided, and an information input / output device 12 such as a personal computer or a smartphone for exchanging information between the host system 13 and a user is connected. The host system 13 includes representative values such as two average values of the first area stored in the frequency distribution information database such as the area information database, for example, the upper section area and the second area, for example, the lower section area. The representative value comparison unit 131 that performs comparison processing (average value, median value, etc.) and the frequency distribution in the upper section area that is the first area (first group) are the second area (second group). A distribution change value for estimating the frequency distribution in the lower section area, for example, a distribution change value calculation unit (conversion equation calculation unit) 132 for calculating a conversion formula is provided, and the upper region that is the first region using the conversion formula or the like. A distribution conversion unit (conversion processing unit) 133 that performs conversion processing of the frequency distribution of the area of the partition and estimates the frequency distribution of the region of the lower section that is the second area, and the estimated lower subsection that is the second area The second region from the frequency distribution of the region A distribution estimation unit (bootstrap processing unit) 134 that performs a distribution estimation process (bootstrap process) for calculating an average value of the entire area of a certain sub-compartment is provided, and the sub-compartment area that is the estimated second region A frequency distribution (unknown value) calculation unit 135 that calculates a frequency distribution such as an unknown attribute value of the sub-region of the second section, which is the second area.

  FIG. 2 is a diagram illustrating an example of a frequency distribution used for information processing according to the present embodiment. FIG. 2A is a frequency distribution table of one group, and shows the relationship between the class and the frequency belonging to it, the total frequency, and the representative value (here, the average value). FIG. 2B is a frequency distribution table of two groups, showing the relationship between the class and the frequency belonging thereto, the total frequency, and the representative value (here, the average value). As an example, as shown in FIG. 8, one group (FIG. 8A) is, for example, a large number of LEDs formed on a semiconductor substrate (one wafer), the class is the emission wavelength band, and the frequency is The number of LEDs having the emission wavelength band on the semiconductor substrate. The second group (FIG. 8B) is, for example, a certain region (for example, the central region) of the LEDs formed on the semiconductor substrate, the class is the emission wavelength band, and the frequency is the center on the semiconductor substrate. This is the number of LEDs having a light emission wavelength band in the region. The class and the frequency can be applied to all characteristics that can create a frequency distribution.

  FIG. 3 is a flowchart showing a flow of information processing according to the present embodiment. First, of the data shown in FIG. 2A and FIG. 2B, the data from y1 ′ to y4 ′ of the two groups is unknown, and the total frequency and the representative value (average value) are known. . It is assumed that these data are stored in the frequency distribution information database 11.

  As shown in FIG. 3, when the process is started (step S1), the frequency distribution g (a) of the first group and the representative values E (a) and E (b) of the first group and the second group are obtained in step S2. Obtained from the frequency distribution information database 11 and input to the host system 13. In step S3, the representative value comparison unit 131 compares E (a) and E (b). If E (a) is substantially equal to (b) (YES in step S3), the frequency distribution (unknown value) The calculation unit 135 estimates the missing information of the second group by applying the frequency distribution of the first group to the frequency distribution of the second group as it is in step S4, and ends the processing (step S6). In the case of No in step S3, the process proceeds to step S7, and the representative value comparison unit 131 determines whether or not E (a)> E (b). In the case of YES, the process proceeds to step S8, and the distribution conversion unit 133 converts the frequency distribution of the second group to the low gradation side by the change value calculated by the distribution change value calculation unit 132. In step S9, a representative value E '(a) is obtained based on the converted frequency distribution. In the case of No in step S7, the process proceeds to step S10, and the frequency distribution of the second group is converted to the high gradation side. In step S9, a representative value E '(a) is obtained based on the converted frequency distribution.

  Next, the processing is repeated by the distribution estimation unit 134. In step S11, E (a) is replaced with E '(a) and g (a) is replaced with g' (a), and the process returns to step S3. When the representative value comparison unit 131 determines that E (a) is substantially equal to (b) in step S3, the process proceeds from step S4 to step S5, where the frequency distribution calculation unit 135 obtains a final frequency distribution, and the process ends. .

As described above, missing information of the second group, that is, y ₁ ′ to y ₄ ′ can be accurately estimated from the frequency distribution of the first group and the representative values of the first and second groups. In the example shown in FIG. 8, since the average value is changed from 330 nm to 360 nm, when the frequency distribution is changed from the first group to the second group, it is converted in a direction in which the distribution is shifted to the long wavelength side.

  According to the present embodiment, even if a missing information part exists in attribute value information in different areas, the missing information part is estimated and used based on one frequency distribution, the attribute average value, and the other attribute average value. The feature analysis can be estimated.

  Next, a more specific example will be described. FIG. 4 (a) shows data obtained by summing up deposit / saving amount information in units of municipalities such as A city and B city corresponding to the upper division area. FIG. 4 (b) shows data obtained by tabulating deposit and saving amount information in units of towns, such as C town of A prefecture B city, which is a subdivision area of A prefecture B city. “-” In FIG. 4 (b) indicates that there is no corresponding data, that is, there is missing information. In other words, the number of corresponding populations is counted by deposit and saving amount in the subordinate towns such as A city, B city and C town. It shows that only the total number of populations in the region and the average value of deposits and savings are counted. In this way, information portions that are not tabulated in the lower-partition area are handled as missing information. Below, the process procedure aiming at the estimation of the number of population according to the deposit and savings amount of A prefecture B city C town which is missing information of this FIG.4 (b) is demonstrated.

  FIG. 5 is a flowchart showing an outline of processing in the host system 13 taking as an example the estimation of the number of corresponding populations by deposit and saving amount of FIG. As an initial setting of processing in the host system 13, first, g (a) is a population distribution according to deposit and saving amount totaled in units of A prefectures and B cities, and h (b) is estimated in units of A prefectures B cities and C towns. The aggregated population distribution by deposit and saving amount is set (step 301). Next, in order to determine the relationship between g (a) and h (b), the average value E (a), which is an example of the representative value of the savings amount of A prefecture B city, and the savings amount of A prefecture B city C town The average value E (b), which is an example of the representative value, is compared (step 302, step 303). When the comparison result of the average value of deposits and savings is E (a) ≒ E (b), it is judged that the shape of population distribution by deposits and savings in A prefecture B city and A prefecture B city C town is almost the same Considering h (b) = g (a), the process proceeds to the next step 310 (step 304). When the comparison result of the average value of the deposit and saving amount is E (a)> E (b), E (a) <E (b), the population distribution g (a) according to the deposit and saving amount of the prefecture A city B is converted. Thus, in order to estimate the population distribution h (b) according to the deposit and saving amount of A city, B city, and C town, the conversion formula f (x) is generated. The conversion formula f (x) shows the relationship between the population distribution g (a) by deposit and saving amount of A prefecture B city and the population distribution h (b) by deposit and saving amount of B prefecture C city of A prefecture, and h ( When b) = g (a), f (x) = x. When the comparison result of the average value of deposits and savings is E (a)> E (b), the population distribution by depositing and savings amount g (a) of A city and B city having a large average value is moved to the negative side as a whole. As a result, the average value is lowered to bring E (a) closer to E (b).

  For this reason, the conversion formula f (x) used in the conversion process from the population distribution g (a) by deposit and saving amount of A prefecture B city to the population distribution h (b) by deposit and saving amount of A prefecture B city C town is f (x ) = Protruded upward compared to x. Based on this, a conversion formula f (x) is generated (step 305). When the comparison result of the average value of deposits and savings is E (a) <E (b), the population distribution by depositing and savings amount g (a) of A city and B city with a small average value is moved to the plus side as a whole. As a result, the average value is raised and E (a) is brought close to E (b). For this reason, the conversion formula f (x) used in the conversion process from the population distribution g (a) by deposit and saving amount of A prefecture B city to the population distribution h (b) by deposit and saving amount of A prefecture B city C town is f (x ) = Lower than x. Based on this, a conversion formula f (x) is generated (step 306).

  FIG. 6 is an image diagram showing the difference in the conversion formula f (x) based on the comparison result between E (a) and E (b). The upper diagram (42) is an image of the conversion equation f (x) convex upward when E (a)> E (b), and the lower diagram (43) is when E (a) <E (b). It is an image of the conversion formula f (x) convex downward. Next, using the generated conversion formula f (x), based on the population distribution g (a) according to the deposit and saving amount of A prefecture B city, the estimated population distribution g according to the deposit and saving amount of A prefecture B city C town '(A) is obtained (step 307). Next, the bootstrap method is used to determine the average value E ′ (a) of the deposit amount of the B city, C city, A prefecture, which is the population from the determined g ′ (a) (step 308). Next, the estimated population distribution g ′ (a) of the A prefecture B city C town by deposit and saving amount is g (a), and the average value E ′ (a) of the deposit amount of A prefecture B city C town is E (a). Then, the evaluation is performed again by comparing the average value E (a) of the deposit and saving amounts of the A city, B city, and C town through the numerical value comparison unit (step 302). When the above processing is repeated to derive E (a) ≈E (b) and the estimation of the population distribution h (b) for each deposit and saving amount in A city, B city and C town is completed (YES in step 302), step 304 is executed. Estimating the population distribution h (b) by deposit / save amount of A prefecture B city / C town through the estimation of the missing information part of the population by deposit / save amount of A prefecture B city / C town (step 310) Exit.

  FIG. 7 is an image diagram of a process of estimating the missing information portion from the calculated population distribution h (b) of the prefecture A city B city C town according to the deposit and saving amount. First, based on the population distribution h (b) 51 for each deposit and saving amount in A prefecture, B city and C town, an estimated population ratio 52 is determined for each deposit and saving amount in A prefecture, B city and C town. Next, the estimated population 54 for each prefecture A city B city C town is calculated by combining the population estimate ratio 52 for each prefecture A city B city C town with the total population 53 of A prefecture B city C town. To do. Through this information processing, the number of corresponding populations by amount of deposit and saving in A prefecture, B city, and C town, which was missing information, is estimated and provided to the user.

  According to the present embodiment, even if the missing information part exists in the attribute value information of the lower partition area, the missing information part based on the frequency distribution of the upper partition area, the attribute average value, and the attribute average value of the lower partition area By estimating and providing to the user, it is possible to perform more detailed regional feature analysis using regional information.

  In the above-described embodiment, the configuration and the like illustrated in the accompanying drawings are not limited to these, and can be changed as appropriate within the scope of the effects of the present invention. In addition, various modifications can be made without departing from the scope of the object of the present invention. Each component of the present invention can be arbitrarily selected, and an invention having a selected configuration is also included in the present invention.

  The present invention can be used as an information processing apparatus.

11 ... Regional information database 12 ... Input / output device 13 ... Host system 131 ... Representative value comparison unit 132 ... Conversion formula calculation unit 133 ... Conversion processing unit 134 ... Bootstrap processing unit 135 ... Unknown value calculation unit 21 ... Deposit and saving information data example 22 ... Deposit and saving information data example 41 of A city, B city and C town ... Fig. 42 of f (x) = x ... Fig. 43 with f (x) = x convex upward ... f (x) = x Fig. 51 with convexity down ... Calculated population distribution by deposit and saving amount of A prefecture B city and C town 52 ... Estimated population ratio by deposit and saving amount of A prefecture B city and C town 53 ... Total population of A prefecture B city and C town 54… Estimated population by amount of deposit and saving in A city, B city and C town

Claims (4)

A representative value comparison unit that compares the first representative value of the first frequency distribution in the first region with the second representative value of the second frequency distribution in the second region different from the first region. When,
In the representative value comparing unit, and a distribution converter for converting the pre-Symbol second frequency distribution in the direction of the small deviation of the distribution of the first frequency distribution and the second frequency distribution,
A new second representative value is obtained based on the converted second frequency distribution converted by the distribution conversion unit, and the new second representative value and the first representative value are obtained as the representative value. compared in comparison unit, having a distribution estimating unit to continue processing the first representative value and the degree of coincidence between the second representative value new said seek the second representative value higher due so An information processing apparatus characterized by the above.   2. The frequency distribution calculating unit according to claim 1, further comprising: a frequency distribution calculating unit configured to calculate a frequency of the second frequency distribution based on the second representative value and the frequency distribution estimated by the distribution estimating unit. Information processing device. An information processing apparatus having a regional information database,
A comparison unit that performs a comparison process of two average values of a region in the upper section and a region in the lower section that are stored in advance in the region information database;
Based on the comparison result, a conversion unit for obtaining a conversion method that approximates the frequency distribution of the upper section area to the frequency distribution of the lower section area;
An estimation unit that performs processing for estimating the frequency distribution of the lower section from the frequency distribution of the upper section area using the conversion method;
An information processing apparatus comprising: a bootstrap processing unit that performs a bootstrap process for obtaining an average value of a population from an estimated frequency distribution. A representative value comparison step of comparing the first representative value of the first frequency distribution in the first region with the second representative value of the second frequency distribution in the second region different from the first region. When,
In the representative value comparison step, the distribution converting step of converting the second frequency distribution before SL direction to reduce the deviation of the distribution of said first frequency distribution second frequency distribution,
A new second representative value is obtained based on the converted second frequency distribution converted in the distribution conversion step, and the new second representative value and the first representative value are used as the representative value. compared in comparison step, to have a distribution estimating step of continuing processing of said first representative value and the degree of coincidence between the second representative value new said seek the second representative value higher due so program for executing information processing how to computer characterized by.

Images