WO2024007094A1 - 一种基于聚类算法和tsvm模型的用户生活缴费分析方法 - Google Patents

一种基于聚类算法和tsvm模型的用户生活缴费分析方法 Download PDF

Info

Publication number
WO2024007094A1
WO2024007094A1 PCT/CN2022/103565 CN2022103565W WO2024007094A1 WO 2024007094 A1 WO2024007094 A1 WO 2024007094A1 CN 2022103565 W CN2022103565 W CN 2022103565W WO 2024007094 A1 WO2024007094 A1 WO 2024007094A1
Authority
WO
WIPO (PCT)
Prior art keywords
payment
information
tsvm
clustering
living
Prior art date
Application number
PCT/CN2022/103565
Other languages
English (en)
French (fr)
Inventor
王晓东
Original Assignee
嘉兴尚坤科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 嘉兴尚坤科技有限公司 filed Critical 嘉兴尚坤科技有限公司
Priority to CN202280018232.2A priority Critical patent/CN116997917A/zh
Priority to PCT/CN2022/103565 priority patent/WO2024007094A1/zh
Publication of WO2024007094A1 publication Critical patent/WO2024007094A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q20/00—Payment architectures, schemes or protocols
    • G06Q20/02—Payment architectures, schemes or protocols involving a neutral party, e.g. certification authority, notary or trusted third party [TTP]

Definitions

  • the invention relates to the field of user living payment management, and in particular to a user living payment analysis method based on a clustering algorithm and a TSVM model.
  • the purpose of the present invention is to provide a user life payment analysis method based on a clustering algorithm and a TSVM model in order to overcome the shortcomings of the above-mentioned existing technologies.
  • a user life payment analysis method based on clustering algorithm and TSVM model including the following steps:
  • S1 Obtain the basic information and living payment information of resident users within the set time and area;
  • S2 Pre-classify the living payment information and obtain pre-classified data divided into two categories;
  • S5 Input the mixed data into the trained TSVM model, and classify and identify the mixed data in two categories according to the optimal number of clustering categories;
  • the basic information and living payment information are obtained through the corresponding business management system and/or information collection system.
  • the basic information includes the user's geographical location information and the user's age information
  • the daily payment information includes payment item information, payment amount information, payment behavior information, payment channel characteristics and payment time-consuming information.
  • the payment channel characteristics include the economic cost coefficient of the payment channel and the collection efficiency of the payment channel.
  • the pre-classification of living payment information specifically includes: using a semi-supervised learning algorithm to annotate the payment behavior information in the living payment information, and the labeling types include online payment and offline payment.
  • step S2 it also includes preprocessing the living payment information, which specifically includes: deleting noise data and living payment information of non-continuous paying users, and normalizing the original living payment information to obtain Standardized payment information data.
  • step S3 the payment item information, payment channel economic cost coefficient, payment time-consuming information, payment amount information and payment channel payment efficiency of the resident users in the set year and set area are used as clustering features to perform clustering. class analysis.
  • the cluster analysis uses the K-Means clustering model, performs cluster analysis on the pre-classified data according to the clustering characteristics, obtains the goodness coefficient of the clustering results, and obtains the optimal number of clustering categories.
  • step S5 the mixed data and the optimal number of clustering categories are used as inputs of the TSVM model to obtain classification recognition results in the two categories respectively.
  • step S5 when the TSVM model performs classification and recognition in step S5, if the distance between the mixed data and each hyperplane divided by the TSVM model is greater than the set threshold, the recognition result will be directly output. Otherwise, the mixed data will be judged to be unqualified and the data will be eliminated. Data classification and identification results.
  • the set threshold is the nearest Euclidean distance from all pre-classified data in the mixed data to the hyperplane.
  • the present invention has the following advantages:
  • the present invention first performs pre-classification through labeling, and uses the TSVM model for classification and identification, effectively utilizing semi-supervised learning, saving the labor cost and labor cost of data set labeling work, and improving the effectiveness and rationality of effective user payment behavior analysis. ;
  • the present invention uses big data mining technology to obtain the basic information and living payment information of residential users within a set time and area, and realizes the characteristic classification of residential users’ living payment behavior through pre-classification, clustering model and TSVM model respectively. Finally, the distribution of payment types and basic information of residential users within the set time and set area is obtained, which provides operational monitoring and business management assistance for relevant payment management departments and enterprise operations, improves the recovery efficiency of living payments and reduces the costs of relevant enterprises. operating costs;
  • payment item information In this invention, payment item information, payment channel economic cost coefficient, payment time-consuming information, payment amount information and payment channel payment efficiency are used as clustering features for cluster analysis, which can reflect the characteristics of resident users in a specific area within a specific period of time. Due to changes in payment behavior habits, the division of payment channels can cover all payment methods of different types of residential users, making the cluster analysis results of payment behavior reasonable, reliable and highly adaptable;
  • the present invention first divides major categories through pre-classification, and then uses the K-Means clustering algorithm to obtain the optimal number of cluster categories, and uses the TSVM model to classify and identify the mixed data, which can dynamically classify and identify the payment behavior of residential users. Improve the reliability and validity of the final results.
  • Figure 1 is a schematic flow chart of the method of the present invention.
  • the present invention provides a user life payment analysis method based on clustering algorithm and TSVM model, which includes the following steps:
  • S1 Obtain the basic information and living payment information of resident users within the set time and area;
  • S2 Pre-classify the living payment information and obtain pre-classified data divided into two categories;
  • S5 Input the mixed data into the trained TSVM model, and classify and identify the mixed data in two categories according to the optimal number of clustering categories;
  • Steps S2, S3 and S5 realize the characteristic classification of residential user payment behavior through pre-classification, clustering model and TSVM model respectively, and finally obtain the distribution of payment types and basic information of residential users within the set time and area.
  • step S4 the pre-classified data and original living payment information are mixed, which can be Under the premise of ensuring the reliability of the original data, the pre-classification results are used to assist in completing the cluster analysis.
  • basic information and living payment information are obtained through the corresponding business management system and/or information collection system, and big data mining technology is used to obtain the basic information and living payment information of resident users in the set time and area.
  • Basic information includes the user's geographical location information and user's age information.
  • Life payment information includes payment item information, payment amount information, payment behavior information, payment channel characteristics and payment time-consuming information.
  • Payment channel characteristics include payment channel economic cost coefficient and payment Channel payment efficiency.
  • pre-categorizing the living payment information specifically includes: using a semi-supervised learning algorithm to annotate the payment behavior information in the living payment information, and the labeling types include online payment and offline payment.
  • cluster analysis can reflect changes in the payment behavior habits of residents in a specific area within a specific period of time. Changes, the division of payment channels can cover all payment methods of different types of residential users.
  • This invention comprehensively considers the economic cost coefficient, payment time and payment amount of payment channels to reflect the changing trend of residential users' payment behavior habits over a period of time; then uses clustering algorithm machine learning to obtain clustering categories and obtain the optimal number of clusters. , and then use the TSVM model to refine the classification and recognition; finally, conduct behavioral analysis and result display based on the classification and recognition.
  • the payment item information in the clustering feature includes electricity, water and gas bills; the economic cost coefficient of the payment channel can well reflect the payment habits of the corresponding residential users in the current year.
  • the payment amount information is the customer's single payment fee, which can reflect the stability of the customer's electricity, water and gas use to a certain extent.
  • payment channels include collection by relevant institutions, payment terminals of various relevant institutions, collection by supermarkets, post offices and convenience stores, collection at bank counters, payment by mobile APPs of relevant institutions, collection by non-financial institutions and collection by financial institutions, and payment by various payment channels. Characteristics, relative costs and repayment efficiency are set based on actual conditions.
  • step S2 it also includes preprocessing the living payment information, which specifically includes: deleting noise data and living payment information of non-continuous paying users, and normalizing the original living payment information, Obtain standardized payment information data.
  • step S3 the payment item information, payment channel economic cost coefficient, payment time-consuming information, payment amount information and payment channel payment efficiency of the resident users in the set year and set area are used as clustering features to perform cluster analysis.
  • the cluster analysis uses the K-Means clustering model to perform cluster analysis on the pre-classified data according to the clustering characteristics, obtain the goodness coefficient of the clustering results, and obtain the optimal number of cluster categories.
  • step S5 the mixed data and the optimal number of clustering categories are used as inputs to the TSVM model, and the classification recognition results in the two categories are obtained respectively.
  • the TSVM model performs classification and recognition, if the distance between the mixed data and each hyperplane divided by the TSVM model is greater than the set threshold, the recognition result will be directly output. Otherwise, the mixed data will be judged to be unqualified and the classification and recognition results of the data will be eliminated.
  • the threshold is set to the nearest Euclidean distance from all pre-classified data in the mixed data to the hyperplane.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Computer Security & Cryptography (AREA)
  • Accounting & Taxation (AREA)
  • Strategic Management (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本发明涉及一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，包括以下步骤：S1：获取设定时间和设定区域内居民用户的基本信息和生活缴费信息；S2：对生活缴费信息进行预分类，得到分为两个类别的预分类数据；S3：对预分类数据进行聚类分析，分别获取两个类别对应的最优聚类类别数量；S4：将预分类数据和原始的生活缴费信息进行混合，得到混合数据；S5：将混合数据输入训练完成的TSVM模型中，根据最优聚类类别数量在两个类别中分别对混合数据进行分类识别；S6：得到设定时间和设定区域内居民用户缴费类型和基本信息的分布情况，与现有技术相比，本发明具有提高对生活缴费行为分析的有效性和合理性等优点。

Description

一种基于聚类算法和TSVM模型的用户生活缴费分析方法 技术领域
本发明涉及用户生活缴费管理领域，尤其是涉及一种基于聚类算法和TSVM模型的用户生活缴费分析方法。
背景技术
在日常生活中，水、电和燃气是我们生活的必需能源，在使用这些能源时会产生相应的费用，需要用户按时向国家相关负责部门进行缴纳。目前用户进行生活缴费主要有两种方式：一是线下缴费。用户使用水费催缴单在小区物业或居委会/村委会缴水费，使用电卡在国家电网营业厅缴电费，使用燃气卡在燃气公司营业厅缴燃气费。二是通过手机或电脑的官方/第三方生活缴费应用缴纳水电燃费用。用户需先在应用中选择服务机构，然后输入卡号/单号查询进行绑定，之后选择/输入缴费金额，最后进行支付操作。
目前随着新型缴费方式的引入，如何制定定制化策略，引导居民用户缴费模式从传统缴费方式向新型缴费方式转变，降低电力/水力/燃气等公司的运营成本，减少生活缴费回收周期，是目前的难点之一，而制定定制化化策略需要首先需要对用户的缴费行为和习惯进行分析，获取不同类型居民用户的不同缴费习惯特征，才能分析不同的类型客户群体的缴费行为特点，归纳总结出不同类型居民用户的特征，为不同类型居民用户提供差异化优质的缴费服务，从而提高生活缴费的回收效率和降低相关企业的运营成本，这是一件非常耗费人工成本的任务，而目前没有有效、直观、合理的方法，对用户的生活缴费行为和习惯进行分析。
发明内容
本发明的目的就是为了克服上述现有技术存在的缺陷而提供一种基于聚类算法和TSVM模型的用户生活缴费分析方法。
本发明的目的可以通过以下技术方案来实现：
一种基于聚类算法和TSVM模型的用户生活缴费分析方法，包括以下步骤：
S1：获取设定时间和设定区域内居民用户的基本信息和生活缴费信息；
S2：对生活缴费信息进行预分类，得到分为两个类别的预分类数据；
S3：对预分类数据进行聚类分析，分别获取两个类别对应的最优聚类类别数量；
S4：将预分类数据和原始的生活缴费信息进行混合，得到混合数据；
S5：将混合数据输入训练完成的TSVM模型中，根据最优聚类类别数量在两个类别中分别对混合数据进行分类识别；
S6：得到设定时间和设定区域内居民用户缴费类型和基本信息的分布情况。
进一步地，所述的基本信息和生活缴费信息通过对应的业务管理系统和/或信息采集系统获取。
进一步地，所述的基本信息包括用户的地理位置信息和用户的年龄信息，所述的生活缴费信息包括缴费项目信息、缴费金额信息、缴费行为信息、缴费渠道特征和缴费耗时信息，所述的缴费渠道特征包括缴费渠道经济成本系数和缴费渠道回款效率。
进一步地，步骤S2中，所述的对生活缴费信息进行预分类具体包括：利用半监督学习算法对生活缴费信息中的缴费行为信息进行标注，标注类型包括线上缴费和线下缴费两种。
更进一步地，所述的步骤S2之前，还包括对生活缴费信息进行预处理，具体包括：删除噪声数据以及非连续缴费用户的生活缴费信息，并对原始生活缴费信息进行归一化处理，得到标准化缴费信息数据。
进一步地，步骤S3中，将设定年份和设定区域内居民用户的缴费项目信息、缴费渠道经济成本系数、缴费耗时信息、缴费金额信息和缴费渠道回款效率作为聚类特征，进行聚类分析。
更进一步地，所述的聚类分析采用K-Means聚类模型，根据聚类特征对预分类数据进行聚类分析，获取聚类结果优度系数，得到最优聚类类别数量。
进一步地，步骤S5中，将混合数据和最优聚类类别数量作为TSVM模型的输入，分别得到两个类别中的分类识别结果。
进一步地，所述的步骤S5中TSVM模型进行分类识别时，若混合数据到TSVM模型划分的各个超平面的距离大于设定阈值，则直接输出识别结果，否则判断该混合数据不合格，剔除该数据的分类识别结果。
更进一步地，所述的设定阈值为混合数据中所有预分类数据到超平面最近的欧式距离。
与现有技术相比，本发明具有以下优点：
1)本发明首先通过标注进行预分类，并且采用TSVM模型进行分类识别，有效利用半监督学习，节省了数据集标注工作的人工成本和人工代价，提高有效用户缴费行为分析的有效性和合理性；
2)本发明利用采用大数据挖掘技术获取设定时间和区域内居民用户的基本信息和生活缴费信息，分别通过预分类、聚类模型和TSVM模型，实现对居民用户生活缴费行为的特征分类，最终得到设定时间和设定区域内居民用户缴费类型和基本信息的分布情况，为相关缴费管理部门和企业运营提供运营监测和业务管理的辅助作用，提高生活缴费的回收效率和降低相关企业的运营成本；
3)本发明中将缴费项目信息、缴费渠道经济成本系数、缴费耗时信息、缴费金额信息和缴费渠道回款效率作为聚类特征进行聚类分析，能够反应特定区域居民用户在特定时间内的缴费行为习惯变化的变化，缴费渠道的划分能够涵盖不同类型居民用户的所有缴费方式，令缴费行为聚类分析结果合理且可靠，适应性高；
4)本发明首先通过预分类进行大类划分，再K-Means聚类算法得到最优聚类类别数量，并对混合数据利用TSVM模型进行分类识别，能够动态对居民用户缴费行为进行分类识别，提高最终结果的可靠性和有效性。
附图说明
图1为本发明方法的流程示意图。
具体实施方式
下面结合附图和具体实施例对本发明进行详细说明。显然，所描述的实施例是本发明的一部分实施例，而不是全部实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动的前提下所获得的所有其他实施例，都应属于本发明保护的范围。
如图1所示，本发明提供一种基于聚类算法和TSVM模型的用户生活缴费分析方法，包括以下步骤：
S1：获取设定时间和设定区域内居民用户的基本信息和生活缴费信息；
S2：对生活缴费信息进行预分类，得到分为两个类别的预分类数据；
S3：对预分类数据进行聚类分析，分别获取两个类别对应的最优聚类类别数量；
S4：将预分类数据和原始的生活缴费信息进行混合，得到混合数据；
S5：将混合数据输入训练完成的TSVM模型中，根据最优聚类类别数量在两个类别中分别对混合数据进行分类识别；
S6：得到设定时间和设定区域内居民用户缴费类型和基本信息的分布情况。
步骤S2、S3和S5分别通过预分类、聚类模型和TSVM模型，实现对居民用户生活缴费行为的特征分类，最终得到设定时间和设定区域内居民用户缴费类型和基本信息的分布情况，为相关缴费管理部门和企业运营提供运营监测和业务管理的辅助作用，提高生活缴费的回收效率和降低相关企业的运营成本，步骤S4中将预分类数据和原始的生活缴费信息进行混合，可以在保证原始数据可靠性的前提下利用预分类结果，辅助完成聚类分析。
其中，基本信息和生活缴费信息通过对应的业务管理系统和/或信息采集系统获取，利用采用大数据挖掘技术获取设定时间和区域内居民用户的基本信息和生活缴费信息。基本信息包括用户的地理位置信息和用户的年龄信息，生活缴费信息包括缴费项目信息、缴费金额信息、缴费行为信息、缴费渠道特征和缴费耗时信息，缴费渠道特征包括缴费渠道经济成本系数和缴费渠道回款效率。具体地，步骤S2中，对生活缴费信息进行预分类具体包括：利用半监督学习算法对生活缴费信息中的缴费行为信息进行标注，标注类型包括线上缴费和线下缴费两种。首先通过标注进行预分类，有效利用半监督学习，节省了数据集标注工作的人工成本和人工代价，提高有效用户缴费行为分析的有效性和合理性。将缴费项目信息、缴费渠道经济成本系数、缴费耗时信息、缴费金额信息和缴费渠道回款效率作为聚类特征进行聚类分析，能够反应特定区域居民用户在特定时间内的缴费行为习惯变化的变化，缴费渠道的划分能够涵盖不同类型居民用户的所有缴费方式。
本发明综合考虑缴费渠道的经济成本系数、缴费耗时和缴费金额，反应居民用户一段时间内的缴费行为习惯变化趋势；然后使用聚类算法机器学习得到聚类类别，获取最优聚类个数，再采用TSVM模型进行分类识别细化；最后根据分类识别进行行为分析和结果展示。
聚类特征中的缴费项目信息包括电费、水费和燃气费；缴费渠道经济成本系数 可以很好的反应当前年份对应居民用户的缴费习惯，缴费渠道经济成本系数越大，表示该用户偏好传统缴费方式且使用频次较高；回款效率越高，表示到账时间越快；缴费耗时信息是指居民用户在收到缴费账单到缴费账单从对应业务系统中销根这一过程所耗费的时间，该指标在一定程度上可以反应用户的缴费积极性和信用度；缴费金额信息为客户的单笔缴费费用，可以从一定程度上反应客户的用电、用水和燃气使用稳定性。另外缴费渠道包括各相关机构坐收、各相关机构缴费终端、超市邮局便利店代收、银行柜台坐收、各相关机构掌上APP缴纳、非金融机构收费和金融机构代收，各缴费渠道的缴费特征、相对成本和回款效率根据实际情况设定。
另外，本实施例中，在步骤S2之前，还包括对生活缴费信息进行预处理，具体包括：删除噪声数据以及非连续缴费用户的生活缴费信息，并对原始生活缴费信息进行归一化处理，得到标准化缴费信息数据。
步骤S3中，将设定年份和设定区域内居民用户的缴费项目信息、缴费渠道经济成本系数、缴费耗时信息、缴费金额信息和缴费渠道回款效率作为聚类特征，进行聚类分析。聚类分析采用K-Means聚类模型，根据聚类特征对预分类数据进行聚类分析，获取聚类结果优度系数，得到最优聚类类别数量。
步骤S5中，将混合数据和最优聚类类别数量作为TSVM模型的输入，分别得到两个类别中的分类识别结果。TSVM模型进行分类识别时，若混合数据到TSVM模型划分的各个超平面的距离大于设定阈值，则直接输出识别结果，否则判断该混合数据不合格，剔除该数据的分类识别结果。其中设定阈值为混合数据中所有预分类数据到超平面最近的欧式距离。
以上所述，仅为本发明的具体实施方式，但本发明的保护范围并不局限于此，任何熟悉本技术领域的工作人员在本发明揭露的技术范围内，可轻易想到各种等效的修改或替换，这些修改或替换都应涵盖在本发明的保护范围之内。因此，本发明的保护范围应以权利要求的保护范围为准。

Claims (10)

  1. 一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，包括以下步骤：
    S1：获取设定时间和设定区域内居民用户的基本信息和生活缴费信息；
    S2：对生活缴费信息进行预分类，得到分为两个类别的预分类数据；
    S3：对预分类数据进行聚类分析，分别获取两个类别对应的最优聚类类别数量；
    S4：将预分类数据和原始的生活缴费信息进行混合，得到混合数据；
    S5：将混合数据输入训练完成的TSVM模型中，根据最优聚类类别数量在两个类别中分别对混合数据进行分类识别；
    S6：得到设定时间和设定区域内居民用户缴费类型和基本信息的分布情况。
  2. 根据权利要求1所述的一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，所述的基本信息和生活缴费信息通过对应的业务管理系统和/或信息采集系统获取。
  3. 根据权利要求1所述的一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，所述的基本信息包括用户的地理位置信息和用户的年龄信息，所述的生活缴费信息包括缴费项目信息、缴费金额信息、缴费行为信息、缴费渠道特征和缴费耗时信息，所述的缴费渠道特征包括缴费渠道经济成本系数和缴费渠道回款效率。
  4. 根据权利要求3所述的一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，步骤S2中，所述的对生活缴费信息进行预分类具体包括：利用半监督学习算法对生活缴费信息中的缴费行为信息进行标注，标注类型包括线上缴费和线下缴费两种。
  5. 根据权利要求4所述的一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，所述的步骤S2之前，还包括对生活缴费信息进行预处理，具体包括：删除噪声数据以及非连续缴费用户的生活缴费信息，并对原始生活缴费信息进行归一化处理，得到标准化缴费信息数据。
  6. 根据权利要求3所述的一种基于聚类算法和TSVM模型的用户生活缴费分 析方法，其特征在于，步骤S3中，将设定年份和设定区域内居民用户的缴费项目信息、缴费渠道经济成本系数、缴费耗时信息、缴费金额信息和缴费渠道回款效率作为聚类特征，进行聚类分析。
  7. 根据权利要求6所述的一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，所述的聚类分析采用K-Means聚类模型，根据聚类特征对预分类数据进行聚类分析，获取聚类结果优度系数，得到最优聚类类别数量。
  8. 根据权利要求1所述的一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，步骤S5中，将混合数据和最优聚类类别数量作为TSVM模型的输入，分别得到两个类别中的分类识别结果。
  9. 根据权利要求8所述的一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，所述的步骤S5中TSVM模型进行分类识别时，若混合数据到TSVM模型划分的各个超平面的距离大于设定阈值，则直接输出识别结果，否则判断该混合数据不合格，剔除该数据的分类识别结果。
  10. 根据权利要求8所述的一种基于聚类算法和TSVM模型的用户生活缴费分析方法，其特征在于，所述的设定阈值为混合数据中所有预分类数据到超平面最近的欧式距离。
PCT/CN2022/103565 2022-07-04 2022-07-04 一种基于聚类算法和tsvm模型的用户生活缴费分析方法 WO2024007094A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202280018232.2A CN116997917A (zh) 2022-07-04 2022-07-04 一种基于聚类算法和tsvm模型的用户生活缴费分析方法
PCT/CN2022/103565 WO2024007094A1 (zh) 2022-07-04 2022-07-04 一种基于聚类算法和tsvm模型的用户生活缴费分析方法

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/103565 WO2024007094A1 (zh) 2022-07-04 2022-07-04 一种基于聚类算法和tsvm模型的用户生活缴费分析方法

Publications (1)

Publication Number Publication Date
WO2024007094A1 true WO2024007094A1 (zh) 2024-01-11

Family

ID=88534342

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/103565 WO2024007094A1 (zh) 2022-07-04 2022-07-04 一种基于聚类算法和tsvm模型的用户生活缴费分析方法

Country Status (2)

Country Link
CN (1) CN116997917A (zh)
WO (1) WO2024007094A1 (zh)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110363071A (zh) * 2019-05-31 2019-10-22 上海海洋大学 一种协同主动学习和直推式支持向量机的海冰检测方法
CN111626614A (zh) * 2020-05-27 2020-09-04 国网上海市电力公司 一种基于电费回收的用户分类方法
CN112258067A (zh) * 2020-10-30 2021-01-22 国网上海市电力公司 基于高斯混合模型聚类算法的低压用户缴费行为分类方法
US10949825B1 (en) * 2017-06-30 2021-03-16 Square, Inc. Adaptive merchant classification

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10949825B1 (en) * 2017-06-30 2021-03-16 Square, Inc. Adaptive merchant classification
CN110363071A (zh) * 2019-05-31 2019-10-22 上海海洋大学 一种协同主动学习和直推式支持向量机的海冰检测方法
CN111626614A (zh) * 2020-05-27 2020-09-04 国网上海市电力公司 一种基于电费回收的用户分类方法
CN112258067A (zh) * 2020-10-30 2021-01-22 国网上海市电力公司 基于高斯混合模型聚类算法的低压用户缴费行为分类方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
DU, HONGLE; ZHANG, YAN: "Transductive Support Vector Machine Algorithm Based on Cluster and Cooperative Labeling", HENAN SCIENCE, vol. 35, no. 1, 31 January 2017 (2017-01-31), pages 22 - 27, XP009552295, ISSN: 1004-3918 *
WANG, LIMEII; LI, JINFENG; YUE, QI: "k Means Clustering based Transductive Support Vector Machine Algorithm", COMPUTER ENGINEERING AND APPLICATIONS, HUABEI JISUAN JISHU YANJIUSUO, CN, vol. 49, no. 14, 31 December 2013 (2013-12-31), CN , pages 144 - 146, XP009552294, ISSN: 1002-8331 *

Also Published As

Publication number Publication date
CN116997917A (zh) 2023-11-03

Similar Documents

Publication Publication Date Title
Xiang et al. Confession or justification: The effects of environmental disclosure on corporate green innovation in China
CN108090823B (zh) 基于SaaS的账务数据管理系统
CN102663650A (zh) 一种企业信用风险分析系统及其使用方法
Peng et al. Digital economy and business investment efficiency: Inhibiting or facilitating?
Liu et al. Coupling coordination degree of government support, financial support and innovation and its impact on economic development
Tomita et al. An algorithm for locating logic design errors
CN110427418A (zh) 一种基于客户能源价值指标体系的客户分析分群方法
CN111932366A (zh) 单证业务处理装置及方法
Xie et al. Understanding the CSR awareness of large construction enterprises in China
CN109102396A (zh) 一种用户信用评级方法、计算机设备及可读介质
Si Construction and application of enterprise internal audit data analysis model based on decision tree algorithm
WO2024007094A1 (zh) 一种基于聚类算法和tsvm模型的用户生活缴费分析方法
CN112258067A (zh) 基于高斯混合模型聚类算法的低压用户缴费行为分类方法
CN112949601A (zh) 一种电子账本关联凭证分析方法、装置及系统
CN112214536B (zh) 一种企业账目数据实时导入管理系统
KR20230103025A (ko) 기업 신용 분석 및 평가 정보 제공 방법, 장치 및 시스템
Ding et al. LSTM Deep Neural Network Based Power Data Credit Tagging Technology.
Jiu-jin et al. Influence of XBRL on the accounting information quality
Sun et al. A new perspective of credit scoring for small and medium-sized enterprises based on invoice data
Li et al. LSTM Deep Neural Network Based Power Data Credit Tagging Technology
CN114155038B (zh) 受疫情影响用户识别方法
Yu Integration of ESG in Financial Institutions: A Study on the Impact of Sustainability Reporting
Zhao Financial Risk Evaluation of Digital Currency Based on CART Algorithm Blockchain
CN114091857A (zh) 财务报表处理方法、装置、电子设备和存储介质
Du The Impact of Financial Geographical Structure on Corporate Pollution Emissions

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 202280018232.2

Country of ref document: CN

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22949691

Country of ref document: EP

Kind code of ref document: A1