JP6257298B2 - Data classification apparatus, data classification method, and data classification program - Google Patents

Description

  The present invention relates to a data classification device, a data classification method, and a data classification program.

  The ATM journal that records the contents of transactions by ATM (Automatic Teller Machine) includes personal information. When this personal information is to be analyzed, it is required to protect privacy.

  Japanese Patent Laid-Open No. 2013-125374 (Patent Document 1) states that “the first of the plurality of records for each type of attribute value of the first attribute that is included in the plurality of records and designated as the object to be obscured. Determining whether the distribution of the number of records satisfies a condition indicating that the bias is large, from the data stored in the data storage unit in which the number of records in which the attribute value of the attribute appears is stored; When the distribution of the number of records satisfies the condition indicating that the deviation is large, the attribute value of the first attribute in at least one of the plurality of records is replaced with the ambiguous data, and the data is stored. Including the process of storing in the part. "

  Japanese Patent Laid-Open No. 2010-72727 (Patent Document 2) states that “the history processing apparatus first classifies each history data included in the user's behavior history data recorded in time series into a history set. Based on one condition, a first process for classifying each history data into several history sets is executed, and after the first process is finished, the history processing apparatus relates to the recording time of the history data and the similarity between the history data. Based on the second condition, a second process is executed in which the isolated data remaining without being classified into the set by the first process is incorporated into any of the sets. "

JP 2013-125374 A JP 2010-72727 A

  With the technique described in Patent Document 1, the privacy of personal information included in the analysis target can be protected by substituting with ambiguous data. However, there is a problem that the amount of information of obfuscated data after replacement is smaller than that of data before replacement.

  In the technique described in Patent Document 2, personal information included in the analysis target can be protected by classifying the history set based on the similarity. Here, the degree of similarity is a numerical value used for processing, etc., and is not a value that has any characteristics or meaning in itself. There was a problem that it was not easy to identify.

  An object of the present invention is to provide a technique that makes it easy to identify on which basis the classified group is classified.

  Of the inventions disclosed in the present application, the outline of typical ones will be briefly described as follows.

  One embodiment of the present invention has an attribute series DB that stores attribute series data obtained by combining static data indicating customer attributes and data accumulated for the customer behavior. A feature extraction unit configured to extract a feature from the attribute series data and add the extracted feature to the attribute series DB in association with a data item as data; In addition, when the data included in the attribute series data is a numerical value, there is a division unit that converts the numerical value into a section. Further, the data processing apparatus includes an output device that displays the data items stored in the attribute series DB and the data in a list in association with each other. Moreover, it has an input device which receives the input which selects the said data item from the said data item which the said output device displays with a list. The input device further includes a grouping unit that groups data extracted based on the data items received by the input device.

  In another embodiment, the feature extraction unit extracts and extracts features from attribute series data obtained by combining static data indicating customer attributes and data accumulated for the customer behavior. A data item adding step of adding the feature as data to the attribute series DB in association with the data item; In addition, when the data included in the attribute series data is a numerical value, the sorting unit includes a numerical value conversion step of converting the numerical value into a section. In addition, the output device includes a display step of displaying the data items stored in the attribute series DB and the data in a list in association with each other. The input device may further include a data item selection step for receiving an input for selecting the data item from the data items displayed by the output device in a list. Further, the grouping unit includes a grouping step for grouping data extracted based on the data items received by the input device.

  In another embodiment, the feature extraction unit extracts and extracts features from attribute series data obtained by combining static data indicating customer attributes and data accumulated for the customer behavior. The computer of the data classification device is caused to execute a data item adding step of adding the feature as data to the attribute series DB in association with the data item. In addition, when the data included in the attribute series data is a numerical value, the sorting unit causes the computer to execute a numerical value conversion step of converting the numerical value into an interval. Further, the output device causes the computer to execute a display step of displaying the data items stored in the attribute series DB in association with the data in a list. Further, the input device causes the computer to execute a data item selection step for receiving an input for selecting the data item from the data items displayed by the output device in a list. Further, the grouping unit causes the computer to execute a grouping step of grouping data extracted based on the data items received by the input device.

  Among the inventions disclosed in the present application, effects obtained by typical ones will be briefly described as follows.

  According to the representative embodiment of the present invention, it is easy to identify on which basis the classified group is classified.

It is a figure which shows the outline | summary of the structural example of the data classification device in one embodiment of this invention. It is a figure which shows the outline | summary of the structural example of the attribute data which attribute DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the structural example of the series data which the series DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the other structural example of the series data which the series DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the structural example of the attribute series data which attribute series DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the whole process in one embodiment of this invention. It is a figure which shows the outline | summary of the feature extraction process in one embodiment of this invention. It is a figure which shows the outline | summary of the other structural example of the attribute series data which attribute series DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the further another structural example of the attribute series data which attribute series DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the further another structural example of the attribute series data which attribute series DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the further another structural example of the attribute series data which attribute series DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the division process in one embodiment of this invention. It is a figure which shows the outline | summary of the further another structural example of the attribute series data which attribute series DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the further another structural example of the attribute series data which attribute series DB in one embodiment of this invention memorize | stores. It is a figure which shows the outline | summary of the item selection process in one embodiment of this invention. It is a figure which shows the outline | summary of the selection screen in one embodiment of this invention. It is a figure which shows the outline | summary of the grouping process in one embodiment of this invention. It is a figure which shows the outline | summary of the structural example of the grouping data which grouping DB in one embodiment of this invention memorize | stores.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that components having the same function are denoted by the same reference symbols throughout the drawings for describing the embodiment, and the repetitive description thereof will be omitted.

<Overall configuration>
FIG. 1 is a diagram showing an outline of a configuration example of a data classification device 1 according to an embodiment of the present invention. As shown in FIG. 1, the data classification device 1 includes an output device 110, an input device 120, an attribute DB 131, a series DB 132, an attribute series DB 133, a grouping DB 134, and a data classification unit 140.

  Further, the data classification device 1 is implemented by predetermined hardware and software. For example, the data classification device 1 includes a processor, a memory, and the like, and a program on the memory executed by the processor causes the computer of the data classification device 1 to function.

  The data classification unit 140 includes a feature extraction unit 141, a sorting unit 142, an item selection unit 143, and a grouping unit 144.

  The attribute DB 131 stores attribute data, which is static data indicating customer attributes such as customer age and sex.

  The series DB 132 stores, for example, series data, which is data accumulated with respect to customer behavior, such as customer purchase logs and customer transaction logs when ATMs are used.

  The attribute series DB 133 stores attribute series data obtained by combining static data indicating customer attributes and data accumulated with respect to customer behavior. The attribute series data is data to be analyzed (hereinafter sometimes referred to as analysis target data).

  The feature extraction unit 141 acquires attribute series data from the attribute series DB 133. The feature extraction unit 141 also acquires the number of elements constituting the feature (data (this data corresponds to data of a data item having the attribute series data)) from the acquired attribute series data, and the adjacent data constituting the data. The ratio of matching elements, the most frequent element that is the most contained in the data, the average value of each element, etc.) are extracted. Further, the feature extraction unit 141 adds the extracted feature as data to the attribute series DB 133 in association with the data item.

  When the data included in the attribute series data is a numerical value, the sorting unit 142 extracts the maximum value and the minimum value of this data. Further, the sorting unit 142 equally divides a section from the extracted minimum value to the maximum value into L (for example, 2). Thereby, the section from the extracted minimum value to the maximum value is divided into L sections. Further, the sorting unit 142 converts the numerical value, which is data, into a section after division. Then, the sorting unit 142 adds the converted data to the attribute series DB 133.

  The output device 110 displays the data items and data stored in the attribute series DB 133 as a list on a selection screen (described later, FIG. 16).

  The input device 120 receives an input for selecting one or more data items from each data item displayed on the selection screen. Thereby, the input device 120 receives an input for selecting a data item for generating a group.

  The grouping unit 144 acquires data corresponding to the data item for which the input device 120 has received an input from the attribute series DB 133. The grouping unit 144 also generates a plurality of groups by combining the acquired data, and classifies the groups based on the group that generated the attribute series data. As a result, the grouping unit 144 groups the data extracted based on the data item that received the input to be selected. The grouping unit 144 also stores the generated classification result for each group in the grouping DB 134.

<Attribute data>
FIG. 2 is a diagram showing an outline of a configuration example of attribute data stored in the attribute DB 131 according to the embodiment of the present invention. The attribute DB 131 stores attribute data, which is static data indicating customer attributes such as customer age and sex. As shown in FIG. 2, the attribute data includes data items such as [ID], [attribute 1 (age)] to [attribute k (gender)]. [ID] indicates a code for identifying a customer. [Attribute 1] to [Attribute k] indicate customer attributes. For example, [Attribute 1] indicates the age of the customer, and [Attribute k] indicates the sex of the customer.

<Series data>
FIG. 3 is a diagram showing an outline of a configuration example of the sequence data stored in the sequence DB 132 according to the embodiment of the present invention. The series DB 132 stores, for example, series data, which is data accumulated with respect to customer behavior, such as customer purchase logs and customer transaction logs when ATMs are used. As shown in FIG. 3, the series data includes data items such as [ID], [Date], [Transaction], and [Store]. [Date] indicates the date on which the transaction was made. [Transaction] indicates the content of the transaction. [Store] indicates the name of the store where the transaction was made.

  As shown in FIG. 4, the series data may be composed of [ID], [year / month], [balance], and the like. [Balance] indicates the balance on a date (for example, the end of the corresponding month) specified from [Year / Month].

<Attribute series data>
FIG. 5 is a diagram showing an outline of a configuration example of attribute series data stored in the attribute series DB 133 according to the embodiment of the present invention. The attribute series DB 133 stores attribute series data obtained by combining static data indicating customer attributes and data accumulated with respect to customer behavior. As shown in FIG. 5, the attribute series data is generated by combining attribute data having the same [ID] and series data. Specifically, in the case of the series data shown in FIG. 3, from the data in an arbitrary range of [date] (for example, 2013/7), the value of [transaction] from the left in ascending order of [date] for each [ID]. A side-by-side list is generated and [2013/7 Transactions] are combined. The attribute series data includes [attribute 1 (age)] to [attribute k (gender)], [series 1 (2013/7 transaction)] to [series m (2013/4 balance, 2013/5 balance, 2013/6 balance). , ...)].

<Overall processing>
FIG. 6 is a diagram showing an overview of the overall processing in one embodiment of the present invention.

  First, in S601, a feature extraction process (described later, FIG. 7) is executed. In the feature extraction process, the feature extraction unit 141 calculates the feature of the data calculated from a plurality of elements (the number of elements constituting the data, the ratio of adjacent elements constituting the data, and the number contained in the data). The most frequent elements, the average value of each element, etc.) are extracted. Then, the feature extraction unit 141 adds data having the extracted feature as a data item to the attribute series DB 133.

  Next, in S602, sorting processing (described later, FIG. 12) is executed. In the classification process, when the data included in the attribute series data is a numerical value, the classification unit 142 extracts the maximum value and the minimum value of the numerical value that is the data. Then, the sorting unit 142 equally divides the extracted section from the minimum value to the maximum value into L (for example, 2). Thereby, the section from the extracted minimum value to the maximum value is divided into L sections. Further, the sorting unit 142 converts each numerical value, which is data, into a section after division. Then, the sorting unit 142 adds the converted data to the attribute series DB 133.

  Next, in S603, an item selection process (described later, FIG. 15) is executed. In the item selection process, the input device 120 accepts an input for selecting a [data item] for generating a group.

  Next, in S604, a grouping process (described later, FIG. 17) is executed. In the grouping process, a group is generated by combining data of each data item that has received an input to be selected in the item selection process. The grouping unit 144 stores the generated groups in the grouping DB 134 (described later, FIG. 18).

<Feature extraction process>
FIG. 7 is a diagram showing an outline of the feature extraction processing in one embodiment of the present invention.

  First, in S701, the feature extraction unit 141 selects one data item from the data items included in the attribute series data stored in the attribute series DB (described above, FIG. 5) 133. Note that the feature extraction unit 141 selects an unselected data item.

  Next, in S702, the feature extraction unit 141 determines whether the data of the data item selected in S701 is a list. Note that the feature extraction unit 141 determines that this data is a list when the data of the data item is composed of a plurality of elements. In S702, when the feature extraction unit 141 determines that the data of the data item is a list (S702-Yes), the process proceeds to S703. On the other hand, if the feature extraction unit 141 determines in S702 that the data item data is not a list (S702-No), the process proceeds to S712. For example, if the data item selected in S701 is [attribute 1 (age)], the feature extraction unit 141 determines that the data is not a list because there is only one element included in the data. On the other hand, when the data item selected in S701 is [Series 1 (2013/7 transaction)], since the data includes a plurality of elements, the feature extraction unit 141 determines that the data is a list. .

  In S703, the feature extraction unit 141 acquires all data of the data item selected in S701 from the attribute series DB 133. For example, when the data item selected in S701 is [series 1 (2013/7 transaction)], the feature extraction unit 141 performs data “payment, payment, transfer”, “payment, payment” as data of this data item. , Payment "," afterglow, afterglow, transfer "," payment, payment, payment ", and" transfer, transfer, afterglow "are acquired from the attribute series DB 133.

  Next, in S704, the feature extraction unit 141 calculates the number of elements constituting each data acquired in S703. For example, the data acquired in S703 is “payment, payment, transfer”, “payment, payment, payment”, “afterglow, afterglow, transfer”, “payment, payment, payment”, “transfer, transfer, afterglow”. In this case, the feature extraction unit 141 calculates the number of elements of each data as “3”.

  In step S <b> 705, the feature extraction unit 141 determines whether all the element numbers calculated in step S <b> 704 match. In S705, when the feature extraction unit 141 determines that all the numbers of elements match (S705-Yes), the data feature can be extracted from the tendency of each element. For this reason, when all the numbers of elements match, the ratio calculated by comparing each element (the ratio is calculated in S707) or the element with the largest number included in the data. In order to extract the frequent element (the most frequent element is extracted in S708) as a feature, the process proceeds to S706. On the other hand, when the feature extraction unit 141 determines in S705 that the number of elements does not match (S705-No), the data feature cannot be extracted from the tendency of each element. Therefore, when the number of elements is not equal, an average value (an average value is calculated in S710) that is an example of a representative value of each element, or an appearance type that excludes duplicate elements in the data ( The appearance type is extracted in step S711), and the process proceeds to step S709.

  In S706, the feature extraction unit 141 determines whether the elements constituting the data acquired in S703 are numerical values. If the feature extraction unit 141 determines in S706 that the element is a numerical value (S706-Yes), the process proceeds to S707. On the other hand, when the feature extraction unit 141 determines in S706 that the element is not a numerical value (S706-No), the process proceeds to S708. For example, the data item selected in S701 is [series m (2013/4 balance, 2013/5 balance, 2013/6 balance,...)], And the data acquired in S703 is “100, 80, ..., 250, 500 "," 100, 120, ..., 100, 150 "," 300, 450, ..., 300, 300 "," 500, 900, ..., 250, 500 "," 120, 48, ..., In the case of “200, 200”, the feature extraction unit 141 determines that the elements constituting the data are numerical values. On the other hand, the data item selected in S701 is [Series 1 (2013/7 transaction)], and the data acquired in S703 includes “payment, payment, transfer”, “payment, payment, payment”, “afterglow, In the case of “afterglow, transfer”, “payment, payment, payment”, and “transfer, transfer, afterglow”, the feature extraction unit 141 determines that the elements constituting the data are not numerical values.

  In S707, the feature extraction unit 141 calculates each ratio by comparing adjacent elements constituting each data acquired in S703. More specifically, when the number of elements is N, the feature extraction unit 141 calculates the ratio of the first element and the second element as the first ratio, and calculates the second element, the third element, Is calculated as the second ratio, and the ratio between the (N-1) th element and the Nth element is calculated as the (N-1) th ratio. Then, each ratio calculated as shown in FIG. 8 is added to the attribute series DB 133. For example, the feature extraction unit 141 calculates the ratio “0.8..., 2.0” for the data “100, 80,..., 250, 500” acquired in S703, and “100, 120,. The ratio “1.2... 1.5” is calculated for “150”, the ratio “1.5... 1.0” is calculated for “300, 450,. .., 250, 500 ”, the ratio“ 1.8..., 2.0 ”is calculated, and“ 120, 48,..., 200, 200 ”is calculated. Then, the feature extraction unit 141 adds each ratio calculated as shown in FIG. 8 to the attribute series DB 133 in association with the data item [series m (ratio)].

  In S708, the feature extraction unit 141 extracts, as the most frequent element, the element having the largest number included in the data among the elements constituting the data acquired in S703. Then, the feature extraction unit 141 adds each mode element extracted as shown in FIG. 9 to the attribute series DB 133 in association with the data item [series 1 (mode element)]. For example, for the data “payment, payment, transfer” acquired in S703, the feature extraction unit 141 extracts “payment” as the most frequent element. For the data “payment, payment, payment”, the feature extraction unit 141 extracts “payment” as the most frequent element. In addition, for the data “afterglow, afterglow, transfer”, the feature extraction unit 141 extracts “afterglow” as the most frequent element. For the data “transfer, transfer, afterglow”, the feature extraction unit 141 extracts “transfer” as the most frequent element. Then, the feature extraction unit 141 adds each mode element extracted as shown in FIG. 9 to the attribute series DB 133. Note that when there are a plurality of mode elements, the feature extraction unit 141 may store each mode element in the attribute series DB 133, or use the last stored element as the mode element. You may make it memorize | store in DB133. After the feature extraction unit 141 adds each extracted most frequent element to the attribute series DB 133, the process proceeds to S712.

  In step S709, the feature extraction unit 141 determines whether the elements constituting the data acquired in step S703 are numerical values. In S709, when the feature extraction unit 141 determines that the element is a numerical value (S709-Yes), the process proceeds to S710. On the other hand, when the feature extraction unit 141 determines in S709 that the element is not a numerical value (S709-No), the process proceeds to S711.

  In S710, the feature extraction unit 141 adds the number of elements calculated in S704 to the attribute series DB 133 in association with the data item [series 2 (number of elements)] as illustrated in FIG. Further, the feature extraction unit 141 calculates an average value of each element, and adds the calculated average value to the attribute series DB 133 in association with the data item [series 2 (average value)] as illustrated in FIG. .

  In S711, the feature extraction unit 141 extracts appearance types by excluding duplicate elements in the data from the elements constituting the data acquired in S703, and the extracted appearance types are shown in FIG. Add to the attribute series DB 133 as shown. Also, the feature extraction unit 141 adds the number of elements calculated in S704 to the attribute series DB 133 as shown in FIG.

  Next, in S712, the feature extraction unit 141 determines whether all data items included in the attribute series data have been selected in S701. If the feature extraction unit 141 determines in S712 that all data items have been selected (S712-Yes), the feature extraction process ends. On the other hand, if the feature extraction unit 141 determines that not all data items have been selected (S712-No), the process proceeds to S701.

<Classification processing>
FIG. 12 is a diagram showing an outline of the sorting process according to the embodiment of the present invention.

  First, in S1201, the classification unit 142 selects one data item from the data items included in the attribute series data. The sorting unit 142 selects an unselected data item.

  Next, in S1202, the sorting unit 142 acquires all data items of the data item selected in S1201. For example, when the data item selected in S1201 is [Series 1 (2013/7 transaction)], the sorting unit 142 displays all the data of this data item as “payment, payment, transfer”, “ “Payment, payment, payment”, “afterglow, afterglow, transfer”, “payment, payment, payment”, “transfer, transfer, afterglow” are acquired from the attribute series DB 133 (described above, FIG. 5).

  Next, in S1203, the sorting unit 142 excludes duplicate data from the data acquired in S1202. Then, the sorting unit 142 calculates the number of data after excluding duplicate data.

  Next, in S1204, the classification unit 142 acquires a threshold value L. It is determined whether the number of data calculated in S1203 exceeds the acquired threshold value L. When the sorting unit 142 determines that the number of data does not exceed the threshold L (S1204-No), the process proceeds to S1212. On the other hand, when the classification unit 142 determines that the number of data exceeds the threshold L (S1204-Yes), the process proceeds to S1205. The threshold value L is described in advance in the setting file. The sorting unit 142 acquires the threshold value L by reading the setting file.

  In S1205, the sorting unit 142 determines whether the data of the data item selected in S1201 is a list. In S1205, when the sorting unit 142 determines that the data item data is a list (S1205-Yes), the process proceeds to S1206. On the other hand, in S1205, when the classification unit 142 determines that the data item data is not a list (No in S1205), the process proceeds to S1209.

  In S1206, the classification unit 142 determines whether the elements constituting the data acquired in S1202 are numerical values. In S1206, when the classification unit 142 determines that the element is a numerical value (S1206-Yes), the process proceeds to S1207. On the other hand, in S1206, when the classification unit 142 determines that the element is not a numerical value (S1206 No), the process proceeds to S1212.

  In S1207, the classification unit 142 extracts, for each data acquired in S1202, the minimum value and the maximum value of each element constituting the data. Then, the sorting unit 142 equally divides the extracted section from the minimum value to the maximum value into L (for example, 2). Thereby, the section from the extracted minimum value to the maximum value is divided into L sections. For example, “1,1,3”, “2,5,4”, “3,4,6”, “2,3,2”, “1,3” which are data of the data item [series j] in S1202 When “1, 6” is acquired from the attribute series DB 133, the sorting unit 142 extracts the minimum value “1” and the maximum value “6”. Then, the sorting unit 142 divides the section “1-6” from the extracted minimum value “1” to the maximum value “6” into two equal parts, thereby obtaining “1 to 3.5” ([1, 3.6 ) And an interval of “3.6 to 6” (corresponding to [3.6, 6], meaning 3.6 to 6) Divide into In this way, by generating a section obtained by dividing the minimum value and the maximum value of each element constituting the data into L pieces, and applying the generated section to each element, the number of divided sections can be reduced to L pieces. Can be suppressed. Therefore, the number of sections can be reduced and the readability can be improved more than dividing the sections for each element.

  Next, in S1208, the classification unit 142 converts each element constituting the data of the data item selected in S1201 into a section including the element after dividing in S1207, and displays the converted data as a diagram. 13 is added to the attribute series DB 133, and the process proceeds to S1212. For example, the sorting unit 142 sets “1” to “1 to 3.5”, “2” to “1 to 3.5”, and “3” that are elements constituting the data of the data item [series j]. To "1-3.5", "4" to "3.6 to 6", "5" to "3.6 to 6", and "6" to "3.6 to 6" The converted data “1 to 3.5, 1 to 3.5, 1 to 3.5”, “1 to 3.5, 3.6 to 6, 3.6 to 6”, “1 to 3 .5, 3.6-6, 3.6-6 "," 1-3.5, 1-3.5, 1-3.5 "," 1-3.5, 1-3.5, 3 .6-6 ”is added to the attribute series DB 133 as shown in FIG.

  In S1209, the classification unit 142 determines whether the data acquired in S1202 is a numerical value. In S1209, when the classification unit 142 determines that the data is a numerical value (S1209-Yes), the process proceeds to S1210. On the other hand, when the classification unit 142 determines in S1209 that the data is not a numerical value (S1209—No), the process proceeds to S1212.

  In S1210, the sorting unit 142 extracts the minimum value and the maximum value of the data for each data acquired in S1202. Then, the sorting unit 142 equally divides the extracted section from the minimum value to the maximum value into L (for example, 2). Thereby, the section from the extracted minimum value to the maximum value is divided into L sections. For example, when “1”, “6”, “4”, “3”, “2”, which is data of the data item [attribute i], is acquired from the attribute series DB 133 in S1202, the sorting unit 142 sets the minimum value to A value “1” and a maximum value “6” are extracted. Then, the dividing unit 142 divides the section “1-6” from the extracted minimum value “1” to the maximum value “6” into two equal parts, so that the sections “1-3” and “4-6” Divide into sections.

  Next, in S1211, the sorting unit 142 converts each data item selected in S1201 into a section including the data after dividing in S1210, and the converted data is shown in FIG. In this manner, the attribute series DB 133 is added, and the process proceeds to S1212. For example, the sorting unit 142 sets the data “1” of the data item [attribute i] to “1-3”, “2” to “1-3”, “3” to “4-6”, “4” ”To“ 4-6 ”,“ 5 ”to“ 4-6 ”, and the converted data“ 1-3 ”,“ 4-6 ”,“ 4-6 ”,“ 4-6 ” , “1-3” are added to the attribute series DB 133 as shown in FIG.

  In S1212, the sorting unit 142 determines whether all data items included in the attribute series data have been selected in S1201. If the sorting unit 142 determines in S1212 that all data items have been selected (Yes in S1212-), the sorting process ends. On the other hand, if the sorting unit 142 determines that all data items have not been selected (S1212-No), the process proceeds to S1201.

<Item selection process>
FIG. 15 is a diagram showing an outline of the item selection processing in the embodiment of the present invention.

  First, in S1501, the item selection unit 143 acquires all the attribute series data stored in the attribute series DB 133.

  Next, in S1502, the item selection unit 143 inputs the attribute series data acquired in S1501 to the output device 110.

  In step S1503, the output device 110 displays a selection screen (described later, FIG. 16) based on the attribute series data input in step S1502. Hereinafter, the selection screen will be described with reference to FIG.

  As shown in FIG. 16, a selection field, [data item], [type], [number of data], [processing], [processing source], and [data] are displayed on the selection screen. Is done. [Data item] indicates the name of the data item. [Type] indicates the type of [Data]. [Type] corresponds to “number”, “number list”, “character”, and the like. “Numerical value” indicates that [data] is composed of one element. “Numeric list” indicates that [data] is composed of a plurality of elements. “Character” indicates that [DATA] is composed of characters. [Number of data] indicates the number of data. [Processing] indicates whether [Data] is converted from a numerical value to a section in the classification process. “Original data” indicates a numerical value before [data] is processed. [Processing source] indicates the name of the data item of the original data before conversion. [Data] indicates the contents of the data.

  The input device 120 receives an input for selecting one or more [data items] from each [data item] displayed on the selection screen. When the input device 120 receives an input for selecting [data item], a check 1601 is displayed in a selection column corresponding to the selected [data item].

  Note that the item selection unit 143 selects the output device 110 so that only one of the [data item] before processing and the [data item] after processing is selected for the [data item] whose [data] is processed. To control.

  That is, the item selection unit 143 selects the selection field corresponding to the post-processing [data item] even if the post-processing [data item] is selected in a state where the pre-processing [data item] is already selected. The output device 110 is controlled so that only the check 1601 is displayed. Specifically, the item selection unit 143 refers to [processing] corresponding to the selected [data item]. Then, when “section” is stored in [processing], the item selection unit 143 refers to the corresponding [processing source]. Then, the item selection unit 143 refers to the [data item] stored in the [processing source], and when [data item] is already selected, the item selection unit 143 displays it in the selection column corresponding to this [data item]. The output device 110 is controlled so that the checked 1601 is deleted. In addition, the item selection unit 143 controls the output device 110 so that a check 1601 is displayed in the selection column corresponding to the newly selected [data item] after processing.

  Moreover, even if the [data item] before processing is selected while the [data item] after processing is already selected, the item selection unit 143 only selects the selection column corresponding to the [data item] before processing. The output device 110 is controlled so that a check 1601 is displayed. For details, refer to [Processing] corresponding to the selected [Data Item]. Then, when “section” is not stored in [processing], the item selection unit 143 stores [data item] stored in the corresponding [processing source] in [processing source] and [processing] ] [Section] is stored, and [data item] with a check 1601 displayed in the selection column is extracted. Then, the item selection unit 143 controls the output device 110 so that the check 1601 displayed in the selection column corresponding to the extracted [data item] is deleted. Further, the item selection unit 143 controls the output device 110 so that a check 1601 is displayed in the selection column corresponding to the newly selected [data item] before processing.

  Further, the output device 110 displays the maximum number of groups 1602. The maximum number of groups 1602 corresponds to the number obtained by multiplying each [data number] of the selected [data item].

  Further, the output device 110 displays a selection completion button 1603. When the input device 120 receives an input for selecting the selection completion button 1603, the item selection process is terminated.

<Grouping process>
FIG. 17 is a diagram showing an outline of the grouping process according to the embodiment of the present invention.

  First, in S1701, the grouping unit 144 generates a group based on the data of the data item that has received the input selected in S1503. Specifically, the grouping unit 144 generates a group by combining data of each data item. For example, if the data items received in S1503 are “attribute i (section)”, [attribute k], and [series j (section)], the data of “attribute i (section)” Are “1 to 3” and “4 to 6”, the data of [attribute k] is “male” and “female”, and the data of [series j (section)] is “1 to 3.5,1”. -3.5, 1-3.5 "and" 1-3.5, 3.6-6, 3.6-6 ". In this case, the grouping unit 144 sets the data “1 to 3” and “4 to 6” of the “attribute i (section)” and the data “male” and “female” of the [attribute k] [series j (section). ] Data “1 to 3.5, 1 to 3.5, 1 to 3.5” and “1 to 3.5, 3.6 to 6, 3.6 to 6” are combined. 1 to 3, male, {1 to 3.5, 1 to 3.5, 1 to 3.5} "and" 1 to 3, female, {1 to 3.5, 1 to 3.5, 1 3.5} ”,“ 1-3, male, {1-3.5, 3.6-6, 3.6-6}} ”,“ 1-3, female, {1-3.5, 3.6-6, 3.6-6} "," 4-6, male, {1-3.5, 1-3.5,1-3.5} "and" 4-6, female , {1-3.5, 1-3.5, 1-3.5} "and" 4-6 men, {1-3.5, 3.6-6, 3.6- A} ", it generates a" 4-6, female, {1～3.5,3.6～6,3.6～6} ".

  Next, in S1702, the grouping unit 144 searches the attribute series DB 133 by using each data included in the group generated in S1701 as a key, thereby obtaining the attribute series data stored in the attribute series DB 133 for each group. Classify. The grouping unit 144 also calculates the number of data matching the key (hereinafter sometimes referred to as a count) for each group. Thereby, the number of data belonging to the group for each group is calculated.

  Next, in S <b> 1703, the grouping unit 144 extracts all the row numbers of all the records extracted in S <b> 1702 (hereinafter may be referred to as a row number list) for each group.

  Next, in S1704, the grouping unit 144 selects one data item from the data items that have received the input selected in S1503. The grouping unit 144 selects an unselected data item.

  Next, in S1705, the grouping unit 144 creates grouped data in which each group generated in S1701, each count calculated in S1702, and each row number extracted in S1703 are associated with each other. Is stored in the grouping DB 134 as shown in FIG.

  Next, in S1706, the grouping unit 144 determines whether the section is after the data of the data item selected in S1704 has been processed. In S1706, when the grouping unit 144 determines that the data is a section (Yes in S1706), the process proceeds to S1707. On the other hand, if the grouping unit 144 determines in step S1706 that the data is not a section (No in step S1706), the process advances to step S1708.

  In S1707, grouping unit 144 acquires a numerical value corresponding to each row number included in the row number list for each group. Then, the grouping unit 144 calculates the minimum value, the maximum value, and the average value of the data items selected in S1704 based on the acquired numerical values. Thereby, the grouping unit 144 calculates a numerical value corresponding to the section. Then, the grouping unit 144 adds the calculated minimum value, maximum value, and average value to the grouping DB 134.

  In S1708, the grouping unit 144 determines whether all data items have been selected in S1704. If the grouping unit 144 determines in S1708 that all data items have been selected (S1708—Yes), the grouping process ends. On the other hand, when the grouping unit 144 determines that not all data items have been selected (No in S1708), the process proceeds to S1704.

<Effects of the present embodiment>
According to the data classification device 1 in the present embodiment described above, the data extracted based on the data item that has received the input selected by the input device 120 is grouped, so that the group after classification is It becomes easy to identify whether it is classified by the standard.

  In addition, by calculating the minimum value, maximum value, and average value, which are numerical values corresponding to the sections, the difference in data can be shown for data having the same section.

  Further, by calculating the number of data belonging to the group for each group, it is possible to perform analysis in which the number of data belonging to the group is used as a weight and the group is handled as weighted data.

  In addition, the characteristics of data can be extracted from the tendency of each element by extracting the ratio calculated by comparing each element or the most frequent element that is the element with the largest number included in the data as the feature. It becomes like this.

  In addition, it is a case where the feature of the data cannot be extracted from the tendency of each element by extracting the average value which is an example of the representative value of each element or the appearance type excluding duplicate elements in the data as the feature. In addition, the feature of the data can be extracted from the representative value of each element.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiment. However, the present invention is not limited to the embodiment, and various modifications can be made without departing from the scope of the invention. Needless to say. For example, data other than attribute series data may be applied as the analysis target data.

DESCRIPTION OF SYMBOLS 1 ... Data classification device, 110 ... Output device, 120 ... Input device, 131 ... Attribute DB, 132 ... Series DB, 133 ... Attribute series DB, 134 ... Grouping DB, 140 ... Data classification unit, 141 ... Feature extraction unit, 142 ... Classification section, 143 ... Item selection section, 144 ... Grouping section, 1601 ... Check, 1602 ... Maximum number of groups, 1603 ... Selection complete button

Claims (9)

An attribute DB that stores customer identification information and customer attributes in association with data items of the respective data;
A series DB for storing the customer identification information, an action history indicating the action of the customer, and a date and time of the action of the customer in association with data items of the respective data;
Based on the identification information of the customer in the attribute DB and the identification information of the customer in the series DB, the customer identification information, the attribute of the customer in the attribute DB, and the action history of the series DB An attribute series DB for storing attribute series data including list data linked in chronological order in a period based on date and time in association with data items of the respective data of the attribute series data ;
The data of the attribute-series data to extract features from the extracted the feature, the feature extraction unit in association with the data item of the data of the feature to add to the attribute lines DB,
When the is an attribute included in the sequence data Lud chromatography data is numeric, a classification unit for converting the numerical into sections,
An output device for displaying a data item stored in the attribute lines DB in the list,
From Lud over data item displays the output device in the list, an input device that receives an input for selecting a data item for generating the group,
And based on the type of data corresponding to the data item that the input device accepts the input, and a grouping unit for grouping attributes series data stored in the attribute lines DB,
I have a,
The feature extraction unit includes, as the feature, a ratio of comparing adjacent elements in an element that is an action history constituting each list data, a most frequent value in an element constituting each list data, and an element constituting each list data A data classification device that calculates at least one of an average value and an appearance type in an element constituting each list data . The data classification device according to claim 1, wherein
A grouping DB for storing results grouped by the grouping unit;
The said grouping part is a data classification apparatus which memorize | stores the classification result for every group in which it grouped in said grouping DB . The data classification device according to claim 2 ,
The sorting unit adds the converted data to the attribute series DB in association with a data item indicating that the data has been converted into sections.
The grouping unit refers to the attribute series DB, generates a group so that customers whose data items that the input device accepts all of the values match are in the same group, and the input device accepts the input. If the data item is a data item indicating that it has been converted into the section, the average value, minimum value, and maximum value are referred to using the numerical values before conversion into the section belonging to the same group with reference to the attribute series DB. Is calculated for each group, the number of attribute series data belonging to the same group is calculated for each group, and the calculated average value, minimum value, maximum value, and number are stored as the classification result in the grouping DB. Classification device. In the data classification device according to any one of claims 1 to 3,
The feature extraction unit, the attribute lines the number of elements in an element that is action history constituting each list data included in the data matches all, if the elements are numeric, compares the adjacent elements as the feature A data classification device that calculates the calculated ratio. In the data classification device according to any one of claims 1 to 4 ,
The feature extraction unit, the attribute-series data element number matches all the elements that are action history constituting each list data contained in the case the element is not numeric, the elements of each list data as the feature extracting the most frequent value in a data classification device. In the data classification device according to any one of claims 1 to 5 ,
The feature extraction unit, the attribute-series data without matching the number of elements in an element that is action history constituting each list data included, if the element is a numerical value, the elements of each list data as the feature A data classification device for calculating an average value in In the data classification device according to any one of claims 1 to 6 ,
The feature extraction unit, the attribute-series data without matching the number of elements in an element that is action history constituting each list data included, if the element is not numeric, the elements of each list data as the feature A data classification device that extracts appearance types. An attribute DB that stores customer identification information and customer attributes in association with data items of the respective data;
A series DB for storing the customer identification information, an action history indicating the action of the customer, and a date and time of the action of the customer in association with data items of the respective data;
Based on the identification information of the customer in the attribute DB and the identification information of the customer in the series DB, the customer identification information, the attribute of the customer in the attribute DB, and the action history of the series DB Data to be executed by a data classification apparatus having an attribute series DB that stores attribute series data including data that is linked in order of date and time in a period based on date and time in association with data items of the respective data of the attribute series data A classification method,
Feature extraction unit, the attribute extracting features from sequence data, the data of the extracted the feature, the data items additional step of adding in association with the data item in the data of the feature in the attribute lines DB,
If classification unit, Lud over data included in the attribute-series data it is numeric, and numeric conversion step of converting the numerical into sections,
Output device, and a display step of displaying the data item stored in the attribute lines DB in the list,
Input device, from Lud over data item displays the output device in the list, a data item selection step of accepting an input to select data items to generate a group,
Grouping unit, a grouping step of the input device is have group Dzu the type of data corresponding to the data item the input of which is accepted, to group attributes series data stored in the attribute lines DB,
I have a,
The data item adding step includes, as the characteristics, a ratio of comparing adjacent elements in elements that are action histories constituting each list data, a most frequent value in elements constituting each list data, and elements constituting each list data Calculating at least one of the average value and the appearance type in the elements constituting each list data,
Data classification method. An attribute DB that stores customer identification information and customer attributes in association with data items of the respective data;
A series DB for storing the customer identification information, an action history indicating the action of the customer, and a date and time of the action of the customer in association with data items of the respective data;
Based on the identification information of the customer in the attribute DB and the identification information of the customer in the series DB, the customer identification information, the attribute of the customer in the attribute DB, and the action history of the series DB A computer of a data classification apparatus having an attribute series DB that stores attribute series data including list data linked in chronological order in a period based on date and time in association with data items of the respective data of the attribute series data. A data classification program to be executed,
Feature extraction unit, the attribute extracting features from sequence data, the data of the extracted the feature, the data items additional step of adding in association with the data item in the data of the feature in the attribute lines DB,
If classification unit, Lud over data included in the attribute-series data is numeric, and numeric conversion step of converting the numerical into sections,
Output device, and a display step of displaying the data item stored in the attribute lines DB in the list,
Input device, from Lud over data item displays the output device in the list, a data item selection step of accepting an input to select data items to generate a group,
Grouping unit, a grouping step of the input device is have group Dzu the type of data corresponding to the data item the input of which is accepted, to group attributes series data stored in the attribute lines DB,
Is executed by the computer of the data classification device ,
The data item adding step includes, as the characteristics, a ratio of comparing adjacent elements in elements that are action histories constituting each list data, a most frequent value in elements constituting each list data, and elements constituting each list data The data classification program which calculates at least one of the appearance value in the element which comprises the average value in each list data, and each list data .

Images