Financial staged consumption recommendation method and system

Abstract

The invention relates to the technical field of financial installment, in particular to a financial installment consumption recommendation method and system. The financial staged consumption recommendation method comprises the following steps: acquiring historical credit score data and historical preference data of a consumer, wherein the historical preference data comprises at least one of living habit data, daily consumption behavior data, future financial plan data and personalized demand data; inputting the historical credit score data and the historical preference data into the initial deep learning model to obtain a target deep learning model; obtaining current credit score data and current preference data of the consumer; and respectively inputting the current credit score data and the current preference data into the target deep learning model to obtain a financial staged consumption recommendation scheme for the consumer. According to the invention, on the basis of the credit score used by the existing financial institution, in combination with the historical data of the consumer, the figure portraying of the consumer is completed so as to recommend the most suitable installment scheme for the consumer.

Description

Technical Field

[0001] The present invention relates to the technical field of financial installment consumption, and in particular to a method and system for recommending financial installment consumption. Background Art

[0002] Financial installment consumption, as a financial tool that allows consumers to pay in installments, has been widely used around the world. This mechanism is designed to allow consumers to purchase goods or services that they need but may not be able to pay for all at once. With the development of digital technology, especially the rapid advancement of Internet finance, mobile payment, big data analysis and artificial intelligence technology, providing consumers with more intimate and reasonable installment consumption solutions has become a hot focus in the financial technology industry.

[0003] Traditionally, financial institutions provide consumers with installment plans through simplified calculation formulas and standardized rules. These plans are mainly based on consumers' credit scores, which are often derived from their repayment records, income status and other standardized financial information. However, this method is relatively one-sided and cannot score consumers' credit in many aspects.

[0004] Therefore, there is an urgent need for a financial installment consumption recommendation method and system to solve the problem that existing credit scores are relatively one-sided. Summary of the invention

[0005] The present invention provides a financial installment consumption recommendation method, which is used to solve the problem that the credit score used in the existing financial installment recommendation is relatively one-sided.

[0006] A first aspect of the present invention provides a financial installment consumption recommendation method, the method comprising:

[0007] Acquire historical credit score data and historical preference data of the consumer, wherein the historical preference data includes at least one of life habit data, daily consumption behavior data, future financial plan data, and personalized demand data;

[0008] Inputting the historical credit score data and the historical preference data into an initial deep learning model respectively to obtain a target deep learning model;

[0009] obtaining current credit score data and current preference data of the consumer;

[0010] The current credit score data and the current preference data are respectively input into the target deep learning model to obtain a financial installment consumption recommendation plan for the consumer.

[0011] In some practicable ways, the step of obtaining the consumer's historical credit score data and historical preference data includes:

[0012] The life habit data, the daily consumption behavior data, the future financial plan data, and the personalized demand data are cleaned and converted respectively to obtain life habit model data, daily consumption behavior model data, future financial plan model data, and personalized demand model data;

[0013] The life habit model data, the daily consumption behavior model data, the future financial plan model data and the personalized demand model data are respectively assigned preset weights to obtain the historical preference data.

[0014] In some practicable methods, the step of inputting the historical credit score data and the historical preference data into an initial deep learning model to obtain a target deep learning model includes:

[0015] Cleaning and converting the historical credit score data to obtain historical credit score model data;

[0016] Assigning a preset weight to the historical credit scoring model data to obtain the historical credit scoring model data with the weight;

[0017] The historical credit scoring model data with weights and the historical preference data are input into the initial deep learning model to obtain the target deep learning model.

[0018] In some practicable methods, the step of inputting the historical credit scoring model data with weights and the historical preference data into the initial deep learning model to obtain the target deep learning model includes:

[0019] Extracting features from the historical credit score model data to obtain weighted historical credit score feature data;

[0020] Extracting features from the historical preference data to obtain weighted historical preference feature data;

[0021] The historical credit score feature data and the historical preference feature data are respectively input into the initial deep learning model to obtain the target deep learning model.

[0022] In some practicable methods, the step of inputting the historical credit scoring model data with weights and the historical preference data into the initial deep learning model to obtain the target deep learning model includes:

[0023] Using a convolutional neural network, extracting features from the historical credit score model data to obtain the historical credit score feature data with weights;

[0024] Using the convolutional neural network, extracting features from the historical preference data to obtain the historical preference feature data with weights;

[0025] The historical credit score feature data and the historical preference feature data are respectively input into the long short-term memory network to obtain the target deep learning model.

[0026] In some implementations, the step of obtaining the consumer's current credit score data and current preference data includes:

[0027] Acquire the credit information category corresponding to the historical credit score data, and the preference information category corresponding to the historical preference data;

[0028] According to the credit information category, obtaining the consumer credit information corresponding to the current credit information category;

[0029] Obtaining the current credit score data according to the consumer credit information corresponding to the current credit information category;

[0030] According to the preference information category, obtaining consumer preference information corresponding to the current preference information category;

[0031] The current preference data is obtained according to the consumer preference information corresponding to the current preference information category.

[0032] In some practicable methods, the step of inputting the current credit score data and the current preference data into the target deep learning model respectively to obtain a financial installment consumption recommendation plan for the consumer includes:

[0033] Acquire the financial installment consumption recommendation plan, and match the financial installment consumption recommendation plan with all feedback results output by the target deep learning model to obtain the financial installment consumption recommendation plan matched with the feedback results;

[0034] Inputting the current credit score data and the current preference data into the target deep learning model respectively to obtain a first feedback result, wherein the first feedback result is one of all the feedback results;

[0035] According to the first feedback result, the financial installment consumption recommendation plan corresponding to the first feedback result is obtained.

[0036] A second aspect of the present invention provides a financial installment consumption recommendation system, which is applied to the aforementioned financial installment consumption recommendation method, and the system comprises:

[0037] A data collection module, used to obtain historical credit score data and historical preference data of consumers, wherein the historical preference data includes at least one of life habit data, daily consumption behavior data, future financial plan data and personalized demand data;

[0038] A deep learning recommendation module, used for inputting the historical credit score data and the historical preference data into an initial deep learning model respectively to obtain a target deep learning model;

[0039] The data collection module is also used to obtain the current credit score data and current preference data of the consumer;

[0040] A feedback module is used to input the current credit score data and the current preference data into the target deep learning model respectively to obtain a financial installment consumption recommendation plan for the consumer.

[0041] A third aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the aforementioned financial installment consumption recommendation method.

[0042] The fourth aspect of the present invention provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor executes the aforementioned financial installment consumption recommendation method.

[0043] Beneficial effects of the present invention:

[0044] A method for recommending financial installment consumption, firstly obtaining the historical credit score data and historical preference data of the consumer, wherein the historical preference data includes at least one of life habit data, daily consumption behavior data, future financial plan data and personalized demand data; then, the historical credit score data and the historical preference data are respectively input into an initial deep learning model to obtain a target deep learning model; secondly, the current credit score data and the current preference data of the consumer are obtained; finally, the current credit score data and the current preference data are respectively input into the target deep learning model to obtain a financial installment consumption recommendation scheme for the consumer. Through the above method, the financial installment consumption recommendation not only considers the credit score of the consumer, but also combines the consumer's preferences. The historical credit score data and the historical preference data of the consumer are respectively input into the trained target deep learning model to form a suitable scheme for the consumer. In addition, the initial deep learning model is trained by the historical credit score data and the historical preference data of the consumer, so as to improve the initial deep learning model and obtain the target deep learning model; a more intelligent, forward-looking and personalized financial installment consumption recommendation method is provided. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 The present invention is a flowchart of a financial installment consumption recommendation method. DETAILED DESCRIPTION

[0047] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0049] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0050] Some names involved in the present invention are explained below to facilitate understanding of the technical solution of the present invention:

[0051] Long Short-Term Memory (LSTM) is an artificial neural network architecture commonly used to process sequence data. Compared with traditional recurrent neural networks (RNNs), LSTM performs better in capturing long-term dependencies.

[0052] Financial institutions provide consumers with installment plans through simplified calculation formulas and standardized rules. However, such plans are usually based only on credit scores and data sources, such as repayment history and income records, which results in incomplete credit score results. In addition, due to the lack of personalized understanding of consumers, the plans provided may not meet their real needs and expectations.

[0053] It should be noted that consumers' financial status, consumption habits, and preferences may change over time, but traditional recommendation systems find it difficult to adapt to these changes.

[0054] In view of these shortcomings, the purpose of the present invention is to provide a more intelligent, forward-looking and personalized financial installment consumption recommendation method and system.

[0055] like Figure 1 As shown, the first aspect of the present invention provides a financial installment consumption recommendation method, the method comprising:

[0056] S100, obtaining historical credit score data and historical preference data of the consumer, where the historical preference data includes at least one of life habit data, daily consumption behavior data, future financial plan data, and personalized demand data.

[0057] Among them, the consumer's historical credit score data can be obtained through conventional or traditional methods, which is the same as the method used by traditional financial institutions to provide installment plans to consumers, and will not be repeated here.

[0058] The historical preference data includes at least one of life habit data, daily consumption behavior data, future financial plan data and personalized demand data.

[0059] Specifically, the life habit data, daily consumption behavior data, future financial plan data and personalized demand data are cleaned and converted respectively, and corresponding life habit model data, daily consumption behavior model data, future financial plan model data and personalized demand model data are obtained.

[0060] Life habit model data, daily consumption behavior model data, future financial plan model data and personalized demand model data are respectively assigned preset weights to obtain historical preference data.

[0061] Among them, data cleaning can remove noise from life habit data, daily consumption behavior data, future financial plan data and personalized demand data. The data cleaning method can be a conventional data cleaning method, and this application does not limit this.

[0062] Data conversion can convert the cleaned data into data that can be used in subsequent steps, such as unifying the format of the data for easy use. The data conversion method is a conventional method, and this application does not limit how to perform data conversion.

[0063] Specifically, since the importance of life habit model data, daily consumption behavior model data, future financial plan model data and personalized demand model data are not the same in actual use, different weights are assigned to life habit model data, daily consumption behavior model data, future financial plan model data and personalized demand model data, respectively, to ensure that when the data is used subsequently, the life habit model data, daily consumption behavior model data, future financial plan model data and personalized demand model data can be clearly used to judge the corresponding financial installment consumption recommendation plan for consumers, so that they can be more accurate.

[0064] The weights assigned to the life habit model data, the daily consumption behavior model data, the future financial plan model data and the personalized demand model data are pre-set and can be adjusted as needed in actual calculations.

[0065] It should be noted that the life habit model data, daily consumption behavior model data, future financial plan model data and personalized demand model data refer to the models formed by profiling people based on historical data. The model formed by profiling is a conventional method, and this application does not limit this. In other words, the aforementioned model in this application refers to the model used to capture and predict the laws of specific phenomena or behaviors in data analysis and machine learning.

[0066] S200, inputting historical credit score data and historical preference data into the initial deep learning model respectively to obtain a target deep learning model.

[0067] Among them, obtaining the target deep learning model includes the following steps:

[0068] Clean and convert historical credit score data to obtain historical credit score model data;

[0069] Assigning preset weights to the historical credit scoring model data to obtain weighted historical credit scoring model data;

[0070] The weighted historical credit scoring model data and historical preference data are input into the initial deep learning model to obtain the target deep learning model.

[0071] Specifically, the data cleaning and conversion of the historical credit score data is the same as the cleaning and conversion method used for the historical preference data, and both adopt conventional methods, which will not be repeated here. After cleaning and converting the historical credit score data, the historical credit score model data is obtained. The historical credit score model data is a kind of data that can be used by the initial deep learning model. In this way, the historical credit score data obtained above can be applied to the initial deep learning model to train the initial deep learning model.

[0072] In addition, since the present invention not only uses historical credit score data, but also uses historical preference data as a reference indicator for the consumer's financial installment consumption recommendation plan, it is necessary to consider the weight of the historical credit score data, thereby giving the historical credit score data a preset weight, that is, the weight given to the historical credit score data is pre-set, so that the historical credit score data has a corresponding weight. Next, the weighted historical credit score model data and the historical preference data (the subcategories included in the historical preference data all have weights, and the historical preference data will have corresponding weights) are input into the initial deep learning model, so that during the training process of the initial deep learning model, the weight of the data is maintained, and finally the target deep learning model is obtained.

[0073] In one embodiment, the steps of inputting the historical credit scoring model data with weights and the historical preference data into the initial deep learning model to obtain the target deep learning model include:

[0074] Extract features from historical credit scoring model data to obtain weighted historical credit scoring feature data;

[0075] Extract features from historical preference data to obtain historical preference feature data with weights;

[0076] The historical credit score feature data and the historical preference feature data are respectively input into the initial deep learning model to obtain the target deep learning model.

[0077] Among them, the historical credit scoring model data and historical preference data were obtained in the above steps. In order to improve the training efficiency of the initial deep learning model, feature extraction was performed on the historical credit scoring model data and historical preference data respectively, so that the data input to the initial deep learning model is representative, thereby reducing the training time of the initial deep learning model.

[0078] Specifically, the historical preference data includes life habit data, daily consumption behavior data, future financial plan data and personalized demand data as follows:

[0079] The life habit data may be the consumer's daily routine and / or exercise habits, etc. Life habits may represent the consumer's long-term habits in daily life.

[0080] For example, the work and rest time may be:

[0081] Average sleep duration: Calculate the consumer’s average daily sleep duration based on their sleep and rest schedule data.

[0082] Wake-up time: Convert the wake-up time into a numerical feature. For example, 8 o’clock in the morning can be represented as the value 8.

[0083] Bedtime: Also convert bedtime into a numerical feature. For example, 11pm can be represented as the value 23.

[0084] Exercise habits:

[0085] Exercise frequency: Based on the consumer’s records or statistics, the number of times a person participates in exercise per week is converted into a feature.

[0086] Exercise duration: The duration of each exercise is counted and converted into the average duration of each exercise.

[0087] Favorite sports type: Encode the consumer’s favorite sports type, such as basketball, running, yoga, etc., and convert it into classification features.

[0088] The correlation between work and rest time and exercise time: You can combine work and rest time with exercise time to analyze whether there is a regular correlation between work and rest time and exercise time, for example, whether people who wake up early are more inclined to exercise in the morning.

[0089] By converting lifestyle data such as work and rest schedules and exercise habits into feature engineering, we can have a more comprehensive understanding of consumers’ lifestyles and behavioral habits, thereby better conducting consumer portrait analysis, personalized recommendations, and evaluating financial installment capabilities.

[0090] Daily consumption behavior data may be a consumer's shopping preferences and / or consumption amounts, etc. Daily consumption behavior may indicate which types of goods a consumer is accustomed to purchasing and the amount of money he or she is willing to spend on such goods.

[0091] For example, shopping preferences may be:

[0092] Popular product categories: Count the product categories that consumers buy most frequently and convert them into one-hot encoding or numerical features.

[0093] Shopping time preference: Analyze consumers’ shopping time preferences during the week, such as weekends or weekdays, and convert them into features.

[0094] Shopping channel preference: record the shopping channels commonly used by consumers, such as online shopping, offline physical stores, etc., and convert them into features.

[0095] The consumption amount can be:

[0096] Average consumption amount: Calculate the average consumption amount of consumers as a numerical feature.

[0097] Maximum consumption amount: Count the consumer’s maximum single consumption amount and convert it into a feature.

[0098] Consumption frequency: The consumption frequency of consumers is counted based on consumption records and converted into features.

[0099] Correlation between shopping preferences and consumption amount: By combining shopping preferences and consumption amount, we can analyze the impact of different shopping preferences on consumption amount and discover consumers' consumption habits and preferences.

[0100] By converting consumer behavior data such as shopping preferences and consumption amounts into feature engineering, we can have a more comprehensive understanding of consumers' consumption behavior characteristics, recommend financial installment plans for consumers, provide personalized recommendations, and provide financial institutions with more accurate risk assessment and service customization.

[0101] The future financial planning model data may be for the consumer's budget planning and / or investment planning, etc.

[0102] Exemplarily, the budget planning and / or investment plan may be:

[0103] Income and expenditure characteristics: The income and expenditure amounts in the budget plan are used as characteristics, which can include monthly or annual total income, total expenditure, and the amount of each category of expenditure (such as living expenses, rent, transportation expenses, etc.).

[0104] You can calculate the difference between income and expenditure as a feature, or calculate indicators such as the proportion of expenditure in total income.

[0105] Savings and investment characteristics: The savings amount, investment amount, investment product type and other information in the investment plan are used as characteristics, which can include monthly or annual savings amount, investment amount, and investment portfolio configuration.

[0106] The ratio of savings to expenditure can be calculated as a feature, or factors such as investment return rate and risk level can be considered.

[0107] Financial goal characteristics: Convert financial goals in budget planning and investment plans into characteristics, such as home purchase plans, retirement plans, etc.

[0108] Time characteristics: Time characteristics such as month, quarter, year, etc. can be introduced to capture the trend of budget planning and investment plans changing over time.

[0109] By extracting and transforming information from budget planning and investment plans into the features required by the initial deep learning model, this information can be better utilized for prediction and analysis to improve the accuracy and generalization ability of the initial deep learning model.

[0110] The personalized demand model data may be the consumer's customized demands and / or special service demands, etc.

[0111] Exemplarily, the customized requirements and / or special service requirements may be:

[0112] Special requirements mark:

[0113] By marking the special or customized demands raised by consumers, they can be converted into binary features to indicate whether special demands exist.

[0114] More detailed tagging can be done based on specific needs, such as mapping different types of special service needs to different features.

[0115] Requirement description conversion:

[0116] Convert consumers’ written or verbal demands into structured features. This can be achieved through natural language processing technologies such as text segmentation and sentiment analysis.

[0117] Convert text descriptions into features such as keywords, sentiment polarity, length, etc. so that machine learning models can better understand and utilize this information.

[0118] Historical demand statistics:

[0119] To analyze the customized demands and special service demands raised by consumers in the past, information such as the number, frequency, and type of historical demands can be used as features.

[0120] By analyzing the changing trends and patterns of historical demand, we can help predict possible special demand in the future.

[0121] The relationship between consumer characteristics and needs:

[0122] The basic characteristics of consumers (such as age, gender, region, occupation, etc.) are correlated with their customized needs and special service needs, and consumer characteristics can be used as input features.

[0123] The correlation between consumer characteristics and different types of demands can be considered to better meet the personalized needs of consumers.

[0124] Demand processing time characteristics:

[0125] Information such as processing time and response time for special service needs can be converted into features to reflect service efficiency and consumer experience.

[0126] By converting customized demands and special service requirements into structured features, the initial deep learning model can better understand consumer needs and provide consumers with personalized services and customized financial installment consumption recommendation plans.

[0127] In the present invention, historical preference data is used to assist historical credit score data, and the two work together to help optimize the target deep learning model.

[0128] In some embodiments, the steps of inputting the weighted historical credit scoring model data and the historical preference data into the initial deep learning model to obtain the target deep learning model include:

[0129] Convolutional neural networks are used to extract features from historical credit scoring model data to obtain weighted historical credit scoring feature data.

[0130] Convolutional neural networks are used to extract features from historical preference data to obtain weighted historical preference feature data.

[0131] The historical credit score feature data and historical preference feature data are respectively input into the long short-term memory network to obtain the target deep learning model.

[0132] Specifically, for the aforementioned historical credit score data and historical preference data, feature extraction is performed using a convolutional neural network, wherein the convolutional neural network is a conventional structure and this application does not improve it. The convolutional neural network is used to complete the feature extraction of the historical credit score model data and the historical preference data, thereby obtaining the corresponding features. In addition, since the original historical credit score model data and the historical preference data have weights, the corresponding features also have corresponding weights. Next, the obtained historical credit score feature data and historical preference feature data are respectively input into the long short-term memory network.

[0133] Long Short-Term Memory (LSTM) is used to perform sequence modeling on consumers' historical data. Sequence modeling is an existing modeling method, and this application does not limit how to perform sequence modeling. Among them, historical data includes historical credit score feature data and historical preference feature data. By analyzing these sequence data, we can better understand consumers' behavior patterns and credit history, thereby improving the accuracy of credit scores. Long short-term memory networks are used to analyze consumers' behavior, preferences, and financial status, and future predictions are made based on this information to recommend the most appropriate installment plan for them.

[0134] It should be noted that both the initial deep learning model and the target deep learning model include two parts: convolutional neural network and long short-term memory network. First, convolutional neural network is used, and then long short-term memory network is used for modeling and analysis, so as to recommend the most suitable installment plan for consumers based on this information. Next, consumers can give feedback on the recommendation results, and the system will use this feedback for further optimization of the model.

[0135] Specifically, consumers will give feedback based on the recommended financial installment consumption plan, and there are two feedback results, agreement or disagreement. For example, when disagreeing, consumers can also provide feedback on the reasons for disagreement. In this way, the consumer's feedback result is given a corresponding weight (pre-set) and used as training data for the target deep learning model. The target deep learning model is trained again to further optimize the target deep learning model.

[0136] S300, obtaining the consumer's current credit score data and current preference data.

[0137] Specifically, obtaining the consumer's current credit score data and current preference data includes the following steps:

[0138] Obtain the credit information category corresponding to the historical credit score data and the preference information category corresponding to the historical preference data.

[0139] According to the credit information category, the consumer credit information corresponding to the current credit information category is obtained.

[0140] The current credit score data is obtained based on the consumer credit information corresponding to the current credit information category.

[0141] According to the preference information category, the consumer preference information corresponding to the current preference information category is obtained.

[0142] According to the consumer preference information corresponding to the current preference information category, the current preference data is obtained.

[0143] It should be noted that the current credit score data and current preference data are data of the consumer within the current period of time, such as the credit score data and current preference data of the consumer within a week or a month, while the historical credit score data and historical preference data can be data of the consumer within 1-5 years. The current credit score data and current preference data are data of the same category as the historical credit score data and historical preference data. For example, the historical preference data includes life habit data, daily consumption behavior data, future financial plan data and personalized demand data. Similarly, the current preference data also includes life habit data, daily consumption behavior data, future financial plan data and personalized demand data, so as to analyze consumers under the same category and form a financial installment consumption recommendation plan.

[0144] S400, input the current credit score data and the current preference data into the target deep learning model respectively to obtain a financial installment consumption recommendation plan for the consumer.

[0145] Specifically, obtaining a financial installment consumption recommendation plan for a consumer includes the following steps:

[0146] Obtain a financial installment consumption recommendation plan, and match the financial installment consumption recommendation plan with all feedback results output by the target deep learning model to obtain a financial installment consumption recommendation plan that matches the feedback results.

[0147] The current credit score data and the current preference data are respectively input into the target deep learning model to obtain a first feedback result, wherein the first feedback result is one of all feedback results.

[0148] According to the first feedback result, a financial installment consumption recommendation plan corresponding to the first feedback result is obtained.

[0149] Specifically, the financial installment consumption recommendation plan is pre-built, and the financial installment consumption recommendation plan is matched with the feedback results output by the target deep learning model. Exemplarily, the range values ​​of all feedback results output by the target deep learning model are in the interval of 1-100, and 1-100 is divided into multiple intervals. For example, every 10 points is an interval, and each interval corresponds to a financial installment consumption recommendation plan. In this way, for example, when the consumer's current credit score data and current preference data are input into the target deep learning model and the result obtained is 81, the corresponding 81-90 interval financial installment consumption recommendation plan will be recommended and displayed to the consumer.

[0150] In summary, the financial installment consumption recommendation method provided by the present invention has the following advantages:

[0151] 1. Comprehensive data collection: Not only considers traditional credit scores, but also combines consumers' daily activities and preferences to provide a more comprehensive data view.

[0152] 2. Highly personalized recommendations: Through deep learning technology, we can gain a deeper understanding of consumers’ needs and preferences, thereby providing them with truly personalized installment plans.

[0153] 3. Dynamic adjustment capability: As consumer behavior changes, recommendations can be adjusted dynamically to ensure that the most appropriate solution is always provided to consumers.

[0154] A second aspect of the present invention provides a financial installment consumption recommendation system, which is applied to the aforementioned financial installment consumption recommendation method, and the system includes:

[0155] A data collection module, used to obtain the consumer's historical credit score data and historical preference data, where the historical preference data includes at least one of life habit data, daily consumption behavior data, future financial plan data, and personalized demand data;

[0156] A deep learning recommendation module is used to input historical credit score data and historical preference data into the initial deep learning model to obtain a target deep learning model;

[0157] The data collection module is also used to obtain the consumer's current credit score data and current preference data;

[0158] The feedback module is used to input the current credit score data and the current preference data into the target deep learning model respectively to obtain the financial installment consumption recommendation plan for consumers.

[0159] The data collection module is responsible for collecting consumer information from various data sources, including but not limited to credit card transactions, bank account activities, online shopping behaviors, social media activities, payrolls, and other financial documents.

[0160] In addition, the deep learning recommendation module also includes data processing and feature engineering modules, which use various data processing techniques to convert raw data into a format that can be used by machine learning models. In addition, feature engineering techniques are used to extract features that are valuable to the recommendation system from raw data.

[0161] Deep learning recommendation model: This module uses deep neural networks, especially convolutional neural networks (CNN) and long short-term memory networks (LSTM), to analyze consumers’ behaviors, preferences, and financial status, and recommend the most suitable installment plan based on this information.

[0162] Feedback module: Consumers can give feedback on the recommendation results, and the system will use this feedback to further optimize the deep learning recommendation model.

[0163] A third aspect of the present invention provides a computer-readable storage medium, which stores computer instructions for causing a computer to execute the aforementioned financial installment consumption recommendation method.

[0164] The fourth aspect of the present invention provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor executes the aforementioned financial installment consumption recommendation method.

[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A financial installment consumption recommendation method, characterized in that: Methods include: Acquire historical credit score data and historical preference data of the consumer, wherein the historical preference data includes at least one of life habit data, daily consumption behavior data, future financial plan data, and personalized demand data; Inputting the historical credit score data and the historical preference data into an initial deep learning model respectively to obtain a target deep learning model; obtaining current credit score data and current preference data of the consumer; The current credit score data and the current preference data are respectively input into the target deep learning model to obtain a financial installment consumption recommendation plan for the consumer.

2. The financial installment consumption recommendation method according to claim 1, characterized in that: The step of obtaining the consumer's historical credit score data and historical preference data includes: The life habit data, the daily consumption behavior data, the future financial plan data, and the personalized demand data are cleaned and converted respectively to obtain life habit model data, daily consumption behavior model data, future financial plan model data, and personalized demand model data; The life habit model data, the daily consumption behavior model data, the future financial plan model data and the personalized demand model data are respectively assigned preset weights to obtain the historical preference data.

3. The financial installment consumption recommendation method according to claim 2, characterized in that: The step of inputting the historical credit score data and the historical preference data into the initial deep learning model to obtain the target deep learning model comprises: Cleaning and converting the historical credit score data to obtain historical credit score model data; Assigning a preset weight to the historical credit scoring model data to obtain the historical credit scoring model data with the weight; The historical credit scoring model data with weights and the historical preference data are input into the initial deep learning model to obtain the target deep learning model.

4. The financial installment consumption recommendation method according to claim 3, characterized in that: The step of inputting the historical credit scoring model data with weights and the historical preference data into the initial deep learning model to obtain the target deep learning model includes: Extracting features from the historical credit score model data to obtain weighted historical credit score feature data; Extracting features from the historical preference data to obtain weighted historical preference feature data; The historical credit score feature data and the historical preference feature data are respectively input into the initial deep learning model to obtain the target deep learning model.

5. The financial installment consumption recommendation method according to claim 4, characterized in that: The step of inputting the historical credit scoring model data with weights and the historical preference data into the initial deep learning model to obtain the target deep learning model includes: Using a convolutional neural network, extracting features from the historical credit score model data to obtain the historical credit score feature data with weights; Using the convolutional neural network, extracting features from the historical preference data to obtain the historical preference feature data with weights; The historical credit score feature data and the historical preference feature data are respectively input into the long short-term memory network to obtain the target deep learning model.

6. The financial installment consumption recommendation method according to claim 5, characterized in that: The step of obtaining the current credit score data and current preference data of the consumer includes: Acquire the credit information category corresponding to the historical credit score data, and the preference information category corresponding to the historical preference data; According to the credit information category, obtaining the consumer credit information corresponding to the current credit information category; Obtaining the current credit score data according to the consumer credit information corresponding to the current credit information category; According to the preference information category, obtaining consumer preference information corresponding to the current preference information category; The current preference data is obtained according to the consumer preference information corresponding to the current preference information category.

7. The financial installment consumption recommendation method according to claim 5, characterized in that: The step of inputting the current credit score data and the current preference data into the target deep learning model to obtain a financial installment consumption recommendation plan for the consumer comprises: Acquire the financial installment consumption recommendation plan, and match the financial installment consumption recommendation plan with all feedback results output by the target deep learning model to obtain the financial installment consumption recommendation plan matched with the feedback results; Inputting the current credit score data and the current preference data into the target deep learning model respectively to obtain a first feedback result, wherein the first feedback result is one of all the feedback results; According to the first feedback result, the financial installment consumption recommendation plan corresponding to the first feedback result is obtained.

8. A financial installment consumption recommendation system, characterized in that: The financial installment consumption recommendation method applied to any one of claims 1 to 7, the system comprising: A data collection module, used to obtain historical credit score data and historical preference data of consumers, wherein the historical preference data includes at least one of life habit data, daily consumption behavior data, future financial plan data and personalized demand data; A deep learning recommendation module, used for inputting the historical credit score data and the historical preference data into an initial deep learning model respectively to obtain a target deep learning model; The data collection module is also used to obtain the current credit score data and current preference data of the consumer; A feedback module is used to input the current credit score data and the current preference data into the target deep learning model respectively to obtain a financial installment consumption recommendation plan for the consumer.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the financial installment consumption recommendation method described in any one of claims 1-7.

10. An electronic device, characterized in that: include: at least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor executes the financial installment consumption recommendation method described in any one of claims 1-7.

Images