TWI669669B - Enterprise grouping method and system - Google Patents

Abstract

An enterprise grouping method is implemented by an enterprise grouping system, and the enterprise grouping system stores a plurality of enterprise materials respectively corresponding to a plurality of enterprises, each enterprise data includes a plurality of remittance information, and each remittance information has a remittance amount, the enterprise The grouping method comprises the following steps: (A) obtaining, for each enterprise data, a total remittance amount based on the remittance information; (B) obtaining, for each enterprise data, a total remittance number based on the remittance information; and (C) The enterprise data is divided into multiple clusters according to the total remittance amount corresponding to each group attribute and the total remittance number corresponding to each enterprise data. The present invention quickly divides similar corporate data into the same group, thereby saving time and labor costs for analyzing the location of the enterprise data in the supply chain one by one.

Description

Enterprise grouping method and system

The invention relates to a data classification method, in particular to an enterprise grouping method and a system thereof.

When financial companies expand their new corporate customers, they often expand their new corporate customers by means of random phone marketing or strange visits, but the success rate of random phone marketing or strange visits is extremely low.

If the financial industry can analyze the old customers' data based on the location of the old customers in the supply chain (such as upstream, midstream, and downstream), the financial industry can analyze the desires based on the old customers' data. Expand new corporate customers to screen and generate a list of potential development customers to increase the success rate of new business customers.

However, the large amount of customer data makes the amount of data huge, and it is difficult to analyze new enterprise customers that are similar to the position of the old customers in the supply chain.

Therefore, the object of the present invention is to provide an analysis that is similar to the old one. The customer's position in the supply chain is intended to expand the enterprise grouping method of new corporate customers.

Therefore, the enterprise grouping method of the present invention is executed by an enterprise grouping system, and the enterprise grouping system stores a plurality of enterprise materials respectively corresponding to a plurality of enterprises, each enterprise data includes a plurality of remittance information, and each remittance information has a remittance amount. The enterprise grouping method comprises a step (A), a step (B), and a step (C).

In this step (A), for each enterprise data, the enterprise grouping system obtains a total remittance amount based on the remittance information.

In the step (B), for each enterprise data, the enterprise grouping system obtains a total number of remittances based on the remittance information.

In the step (C), the enterprise grouping system divides the enterprise data into multiple clusters according to the total remittance amount and the total remittance amount corresponding to each group attribute corresponding to each enterprise data.

Another object of the present invention is to provide an enterprise grouping system that can analyze new enterprise customers that are similar to the location of the old customer in the supply chain.

Thus, the enterprise grouping system of the present invention comprises a storage unit and a processing unit.

The storage unit stores a plurality of enterprise materials respectively corresponding to a plurality of enterprises, each enterprise information includes a plurality of remittance information, and each remittance information has a remittance amount.

The processing unit is electrically connected to the storage unit, and for each enterprise data, the processing unit obtains a total remittance amount according to the remittance information, and for each enterprise capital The processing unit obtains a total remittance number according to the remittance information, and the processing unit divides the enterprise data into a plurality of clusters according to the total remittance amount and the total remittance amount corresponding to each group attribute corresponding to each enterprise data.

The effect of the invention is that the processing unit quickly divides similar enterprise data into the same group according to the grouping attributes corresponding to each enterprise data, thereby analyzing the desire to expand similar to the position of the old customer in the supply chain. New corporate customers.

100‧‧‧Enterprise Grouping System

11‧‧‧ storage unit

12‧‧‧Processing unit

201~208‧‧‧Steps

31~39‧‧‧Steps

Other features and advantages of the present invention will be apparent from the embodiments of the present invention, wherein: FIG. 1 is a block diagram illustrating an embodiment of the enterprise grouping system of the present invention; FIG. 2 is a flow chart illustrating An embodiment of the enterprise grouping method of the present invention; and FIG. 3 is a flowchart to assist in explaining the sub-steps included in step 208 of FIG.

Referring to FIG. 1, an embodiment of the enterprise grouping system 100 of the present invention includes a storage unit 11 and a processing unit 12 electrically connected to the storage unit 11. The storage unit 11 is, for example, a hard disk drive (HDD), and the processing unit 12 is, for example, a central processing unit (CPU).

The storage unit 11 stores a plurality of enterprise materials respectively corresponding to a plurality of enterprises. Each enterprise information includes multiple remittance information, and each remittance information has a remittance amount and a remittance time. It is worth noting that these enterprises are enterprises of a user's customer or enterprises that are not customers of the user and have a remittance record with the customer's customers. These enterprises are the same industrial ecological chain, such as steel wire. Industry, sheet steel industry, steel rolling and extrusion industry, strip steel industry, etc.

Referring to FIG. 1 and FIG. 2, an embodiment of how the enterprise grouping system 100 of the present invention performs the enterprise grouping method of the present invention will be described. The steps included in the embodiment will be described below.

In step 201, for each enterprise data, the processing unit 12 obtains a total remittance amount based on the remittance amounts of the remittance information, that is, the sum of the remittance amounts of the remittance information.

In step 202, for each enterprise data, the processing unit 12 obtains a total number of remittances based on the remittance information, that is, the number of remittance information.

In step 203, for each enterprise data, the processing unit 12 obtains an average amount of remittances based on the remittance amounts of the remittance information, that is, an average of the remittance amounts of the remittance information.

In step 204, for each enterprise data, the processing unit 12 obtains a remittance amount variation σ ² according to the remittance amounts of the remittance information, and the remittance amount variation σ ² can be expressed by the following formula: Where N is the total number of remittances, x _i is the remittance amount of the i-th remittance information, and μ is the average amount of the remittance.

In step 205, for each enterprise data, the processing unit 12 obtains the number of remittances within a predetermined period according to the remittance time of the remittance information. It is noted that the processing unit 12 is configured by statistical correspondence. The number of remittance information with the remittance time in the predetermined period is obtained to obtain the number of remittances.

In step 206, for each enterprise data, the processing unit 12 obtains an average remittance interval number according to the remittance time of the remittance information, that is, an average of the number of days between two adjacent remittance times. For example, the company information has 3 remittance information. The remittance time of the remittance information is January 31, February 28, and March 31, and the average remittance interval is (28+31). = 29.5 days.

In step 207, for each enterprise profile, the processing unit 12 obtains a self-correlation verification value indicating the randomness of the enterprise by using the Ljung-Box Q test algorithm according to the remittance time of the remittance information.

It should be noted that, in other embodiments of the present invention, the execution of steps 201-207 may also be performed separately and without sequential order, and is not limited thereto.

In step 208, the processing unit 12 corresponds to each enterprise data. The total remittance amount as the plurality of group attributes, the total remittance number, the remittance average amount, the remittance amount variation, the remittance number, the average remittance interval days, and the self-correlation verification value, using a clustering algorithm Divide these corporate data into multiple clusters. It should be noted that in the present embodiment, the processing unit 12 divides the enterprise data into three clusters, for example.

It is to be noted that, in other embodiments, the processing unit 12 may use the grouping algorithm to divide the enterprise data into the clusters according to the total amount of the remittance corresponding to each enterprise data and the total number of remittances. After the steps 201 and 202 are executed, the step 208 is directly executed, or the average amount of the remittance, the variance of the remittance amount, and the remittance are further calculated according to the total remittance amount corresponding to each enterprise data and the total remittance number. The number of times, the average number of remittance intervals, and the self-related verification value are used to divide the enterprise data into the clusters by using the clustering algorithm, that is, after performing at least one of steps 201, 202, and 203 to 207, Step 208 is directly executed.

Referring to FIG. 3, in the embodiment, the processing unit 12 divides the enterprise data into the clusters by using a k-means grouping algorithm. The following describes the sub-steps 31-39 included in step 208, namely k-means. The steps of the clustering algorithm.

In step 31, the processing unit 12 randomly selects a plurality of enterprise materials respectively as the initial centers of the clusters from the enterprise materials.

In step 32, for each non-initial central enterprise data that is not the initial center in the enterprise materials, the processing unit 12 is based on the non-initial center enterprise resources. The similarity between the non-initial central enterprise data and each initial center is calculated by the corresponding grouping attributes corresponding to the materials and the corresponding grouping attributes corresponding to each initial center. For example, the total remittance amount corresponding to the non-initial central enterprise data, the total remittance amount, the remittance amount variation, the remittance number, the average remittance interval, and the self The total amount of the remittance corresponding to the enterprise data of the initial center of each cluster, the total remittance amount, the average amount of the remittance, the variance of the remittance amount, the number of remittances, the average remittance interval, and the self The relevant verification values are subtracted to calculate the similarity between the non-initial central enterprise data and the enterprise data of the initial center of each cluster.

In step 33, for each non-initial central enterprise data, the processing unit 12 obtains, from the initial centers, a non-initial central enterprise data according to the similarity between the non-initial central enterprise data and each initial center. The target initial center with the highest similarity, and classifies the non-initial center enterprise data into the cluster corresponding to the target initial center. For example, there are three centers of A, B, C, and the difference between the total amount of remittances corresponding to a non-initial center enterprise data and the total number of remittances is the lowest among the three, and the remittance corresponding to the non-initial center enterprise data. The difference between the amount of variance is the lowest among the three, the difference between the number of remittances corresponding to the non-initial center enterprise data, the difference between the average remittance interval days, and the difference between the self-relevant verification values is the lowest among the three, since the C center has The lowest is the lowest, so the C center has the highest similarity with the non-initial center enterprise data and is the initial center of the target.

In step 34, for each cluster corresponding to the initial center, the processing unit 12 re-acquires the center of the cluster and its corresponding plurality according to the grouping attributes corresponding to each enterprise data classified into the cluster. Baseline grouping attribute. It should be noted that each enterprise data classified into the cluster includes enterprise data as an initial center of the clusters. In this embodiment, the reference cluster attributes are corresponding to each enterprise data in the cluster. The average of the equal group attributes, that is, the average of the total remittance amounts corresponding to all enterprise data in the cluster, the average of the total remittances, the average of the remittance average amounts, and the average of the remittance amount variances. The average of the number of such remittances, the average of the average remittance intervals, and the average of the self-correlation tests.

In step 35, for each enterprise data, the processing unit 12 calculates the similarity between the enterprise data and each center according to the grouping attributes corresponding to the enterprise data and the benchmark group attributes corresponding to each center. degree.

In step 36, for each enterprise data, the processing unit 12 obtains, from the centers, a target center having the highest similarity with the enterprise data according to the similarity between the enterprise data and each center, and Classify the enterprise data into the cluster corresponding to the target center.

In step 37, for each cluster corresponding to the center, the processing unit 12 determines whether each enterprise data classified into the cluster corresponding to the center and each enterprise data classified into the cluster corresponding to the initial center It's exactly the same. When the process When the unit 12 determines that each enterprise data classified into the cluster corresponding to the center is not exactly the same as each enterprise data classified into the cluster corresponding to the initial center, the process proceeds to step 38; when the processing unit 12 determines When each enterprise data classified into the cluster corresponding to the center is identical to each enterprise data classified into the cluster corresponding to the initial center, that is, when the cluster members are no longer changed, the flow proceeds to step 39.

In step 38, for each cluster corresponding to the center, the center is taken as the initial center, and step 34 to step 37 are re-executed.

In step 39, for each cluster corresponding to the center, the cluster corresponding to each center is obtained.

It should be noted that, in this embodiment, after the processing unit 12 obtains three clusters, the processing unit 12 can further analyze the location of the cluster in the supply chain according to the enterprise data of each cluster, for example, each cluster. When the average remittance interval of an enterprise data is low and the average remittance amount is low, the cluster is downstream. If the average remittance amount of each enterprise data of the cluster is high, the cluster may be upstream, so that the user is expanding the business. According to the position of each company in the supply chain, it can judge the business and financial products suitable for expansion.

In summary, the enterprise grouping method and system thereof of the present invention, by the processing unit 12, quickly classify similar enterprise data into the same group according to the grouping attributes corresponding to each enterprise data, thereby analyzing similarities to the old ones. Customer desire in the position of the supply chain The purpose of the present invention is indeed achieved by expanding new corporate customers.

However, the above is only the embodiment of the present invention, and the scope of the invention is not limited thereto, and all the simple equivalent changes and modifications according to the scope of the patent application and the patent specification of the present invention are still Within the scope of the invention patent.

Claims (14)

An enterprise grouping method is implemented by an enterprise grouping system, and the enterprise grouping system stores a plurality of enterprise materials respectively corresponding to a plurality of enterprises, each enterprise information includes a plurality of remittance information, and each remittance information has a remittance amount and a remittance Time, the enterprise grouping method comprises the following steps: (A) obtaining, for each enterprise data, a total remittance amount according to the remittance information; (B) obtaining, for each enterprise data, a total remittance number according to the remittance information; (C) for each enterprise data, obtain the number of remittances within a predetermined period of time based on the remittance information; (D) the total remittance amount corresponding to each of the plurality of subgroup attributes corresponding to each enterprise data, the total remittance The number of times and the number of remittances, the enterprise data is divided into multiple clusters; and (E) for each cluster, the location of the cluster in the supply chain is analyzed based on the enterprise data of the cluster. The enterprise grouping method according to claim 1, wherein the step (D) comprises the following steps: (D-1) randomly selecting, from the enterprise materials, a plurality of enterprise materials respectively as initial centers of the clusters. (D-2) for each non-initial central enterprise data of the non-initial center of the enterprise data, according to the grouping attributes corresponding to the non-initial center enterprise data and the grouping attributes corresponding to each initial center , classifying the non-initial central enterprise data into one of the clusters (D-3) for each cluster corresponding to the initial center, reacquiring the center of the cluster and its corresponding multiple reference groups according to the group attributes corresponding to each enterprise data classified into the cluster Attribute; (D-4) classifying the enterprise data into one of the clusters for each enterprise data based on the grouping attributes corresponding to the enterprise data and the benchmark group attributes corresponding to each center (D-5) For each cluster corresponding to the center, it is determined whether each enterprise data classified into the cluster corresponding to the center is identical to each enterprise data classified into the cluster corresponding to the initial center; (D-6) For each cluster corresponding to the center, when it is determined that each enterprise data classified into the cluster corresponding to the center is not exactly the same as each enterprise data classified into the cluster corresponding to the initial center Taking the center as the initial center and re-executing the step (D-3) to the step (D-5); and (D-7) for the cluster corresponding to each center, when the determination is classified to the center Each enterprise of the corresponding cluster The initial data center is classified corresponding to each enterprise cluster information is the same, each center to obtain the corresponding cluster. The enterprise grouping method according to claim 2, wherein the step (D-2) comprises the following steps: (D-2-1) for each non-initial central enterprise data, according to the non-initial central enterprise data corresponding to Calculate the non-initial center enterprise data and each of the equal group attributes and the corresponding group attributes corresponding to each initial center (D-2-2) For each non-initial central enterprise data, based on the similarity between the non-initial central enterprise data and each initial center, obtain one from the initial centers The non-initial central enterprise data corresponds to the target initial center with the highest similarity, and the non-initial central enterprise data is classified into the cluster corresponding to the target initial center; the step (D-4) includes the following steps: (D-4- 1) For each enterprise data, calculate the similarity between the enterprise data and each center according to the grouping attributes corresponding to the enterprise data and the benchmark group attributes corresponding to each center; and (D-4) -2) For each enterprise data, according to the similarity between the enterprise data and each center, obtain a target center with the highest similarity corresponding to the enterprise data from the centers, and classify the enterprise data into The target center corresponds to the cluster. The enterprise grouping method according to claim 1, further comprising the following steps before the step (D): (F) obtaining, for each enterprise data, a remittance amount variation amount according to the remittance information; wherein, in step (D) According to the total remittance amount and the total remittance number, the enterprise data is further divided into the clusters according to the variance of the remittance amount. For the enterprise grouping method described in claim 1, each remittance information further has a remittance time, and before step (D), the following steps are further included: (G) for each enterprise data, obtaining a flat according to the remittance information The average remittance interval days; wherein, in step (D), the enterprise data is further divided into the clusters according to the total remittance amount and the total remittance number according to the average remittance interval days. The enterprise grouping method according to claim 1, before the step (D), further comprising the following steps: (H) obtaining, for each enterprise data, an average amount of the remittance according to the remittance information; wherein, in step (D) In the meantime, according to the total remittance amount and the total remittance number, the enterprise data is further divided into the clusters according to the average amount of the remittance. For the enterprise grouping method described in claim 1, each remittance information further has a remittance time. Before the step (D), the following steps are further included: (I) for each enterprise data, use Ljung according to the remittance information. The -Box Q test algorithm obtains a self-correlation verification value indicating the randomness of the enterprise; wherein, in step (D), the self-related verification value is further determined not only according to the total remittance amount and the total remittance number. Divide these corporate data into these clusters. An enterprise grouping system comprises: a storage unit storing a plurality of enterprise materials respectively corresponding to a plurality of enterprises, each enterprise information comprising a plurality of remittance information, each remittance information having a remittance amount and a remittance time; and a processing unit , electrically connected to the storage unit, for each enterprise The processing unit obtains a total remittance amount according to the remittance information, and for each enterprise data, obtains a total remittance number according to the remittance information, and for each enterprise data, the processing unit obtains the remittance information according to the remittance information. For a number of remittances within a predetermined period of time, the processing unit divides the enterprise data into multiple clusters according to the total remittance amount, the total remittances, and the number of remittances corresponding to each of the plurality of group attributes corresponding to each enterprise data. Finally, the location of the cluster in the supply chain is further analyzed based on the enterprise data of each cluster. The enterprise grouping system of claim 8, wherein the processing unit randomly selects, from the enterprise materials, a plurality of enterprise materials respectively as initial centers of the clusters, and for each of the enterprise materials The non-initial central enterprise data that is not the initial center, the processing unit classifies the non-initial central enterprise data into the clusters according to the grouping attributes corresponding to the non-initial center enterprise data and the grouping attributes corresponding to each initial center One of the clusters, and for each cluster corresponding to the initial center, the processing unit reacquires the center of the cluster and its corresponding plurality according to the group attribute corresponding to each enterprise data classified into the cluster. Baseline grouping attribute. The enterprise grouping system of claim 9, wherein, for each non-initial central enterprise data, the processing unit is configured according to the grouping attribute corresponding to the non-initial central enterprise data and the grouping attribute corresponding to each initial center Calculating the similarity between the non-initial central enterprise data and each initial center, and for each non-initial central enterprise data, the processing unit is based on the similarity between the non-initial central enterprise data and each initial center. In the initial center, a target initial center having the highest similarity with the non-initial center enterprise data is obtained, and the non-initial center enterprise data is classified into a cluster corresponding to the target initial center, and the processing is performed for each enterprise data. The unit calculates the similarity between the enterprise data and each center according to the group attribute corresponding to the enterprise data and the benchmark group attributes corresponding to each center, and for each enterprise data, the processing unit according to the The similarity between the enterprise data and each center, from which a target center having the highest similarity with the enterprise data is obtained, and the enterprise data is classified into a cluster corresponding to the target center. The enterprise grouping system according to claim 8, wherein, for each enterprise data, the processing unit obtains a remittance amount variation number according to the remittance information, and the processing unit not only according to the total remittance amount and the total remittance amount Further, according to the variance of the remittance amount, the enterprise data is divided into the clusters. The enterprise grouping system of claim 8, wherein each remittance information stored by the storage unit further has a remittance time, and for each enterprise data, the processing unit obtains an average remittance interval number according to the remittance information. The processing unit further divides the enterprise data into the clusters according to the total remittance amount and the total remittance number according to the average remittance interval days. The enterprise grouping system of claim 8, wherein, for each enterprise data, the processing unit obtains an average amount of remittance according to the remittance information, and the processing unit further bases not only according to the total remittance amount and the total remittance amount The average amount of the remittance, dividing the corporate data into these groups set. The enterprise grouping system of claim 8, wherein each remittance information stored by the storage unit further has a remittance time, and for each enterprise data, the processing unit uses the Ljung-Box Q test calculation according to the remittance information. The method obtains a self-correlation verification value indicating the randomness of the enterprise, and the processing unit further divides the enterprise data into the clusters according to the total remittance amount and the total remittance number according to the self-related verification value.