Loan interest rate prediction method and device, budgeting system and equipment, and computer program product

Abstract

The invention discloses a loan interest rate prediction method and device, a budgeting system and equipment, and a computer program product. The loan interest rate prediction method comprises the following steps: obtaining loan interest rate influence variable data corresponding to a loan product according to the type of the loan product; determining a loan interest rate prediction model corresponding to the loan product according to the type of the loan product; and according to the loan interest rate influence variable data corresponding to the loan product, performing loan interest rate prediction by using the loan interest rate prediction model corresponding to the loan product to obtain a loan interest rate prediction result. According to the loan interest rate prediction method provided by the embodiment of the invention, product characteristics and data composition of different loans are analyzed, and different types of loan interest rate prediction models are adopted for prediction, so that the models and the data play a maximum role, and the accuracy and reliability of loan interest rate prediction are improved.

Description

Technical Field

[0001] This application relates to the technical field of loan interest rate prediction, and particularly to a loan interest rate prediction method, device, budget system, equipment, and computer program product. Background Art

[0002] In the financial field, especially in interest rate prediction, the existing technologies face various challenges and limitations. With the continuous development and complexity of the financial market, higher requirements are put forward for the scientific nature, accuracy, and dynamic adaptability of interest rate prediction. However, the current technical means still have deficiencies in these aspects.

[0003] First, the traditional budget calculation method lacks scientific nature. With the gradual improvement of the requirements for comprehensive budget management and refined budgeting, the budget calculation has gradually shifted from relying on experience to being data-driven. However, the traditional budget calculation method often relies on subjective judgment and lacks rigorous model support, making it difficult to achieve scientific and systematic prediction. This method not only limits the accuracy of the budget but also makes it difficult to cope with complex business scenarios, and it is difficult to meet the needs of modern budget management for dynamic data support and accurate decision-making.

[0004] Second, there are obvious shortcomings in the dynamic adaptability of the existing technologies. The rapid changes in the external economic environment put forward higher requirements for the dynamic adaptability of the budget model. However, most of the existing technologies rely on single time series analysis methods or simple factor addition methods, making it difficult to effectively cope with diverse scenarios. Especially in the context of complex economic fluctuations, the comprehensiveness and reliability of the prediction results are insufficient, and it is difficult to provide stable decision-making support for enterprises. Summary of the Invention

[0005] Embodiments of this application provide a loan interest rate prediction method, device, budget system, equipment, and computer program product to improve the accuracy of loan interest rate prediction.

[0006] Embodiments of this application adopt the following technical solutions:

[0007] In a first aspect, embodiments of this application provide a loan interest rate prediction method, and the loan interest rate prediction method includes:

[0008] Obtain the loan interest rate influence variable data corresponding to the loan product according to the type of the loan product;

[0009] Determine the loan interest rate prediction model corresponding to the loan product according to the type of the loan product;

[0010] Perform loan interest rate prediction using the loan interest rate prediction model corresponding to the loan product according to the loan interest rate influence variable data corresponding to the loan product to obtain a loan interest rate prediction result.

[0011] Optionally, the loan interest rate prediction model corresponding to the loan product determined according to the type of the loan product includes:

[0012] If the type of the loan product is personal loan, determine that the loan interest rate prediction model corresponding to the loan product is the personal loan interest rate prediction model;

[0013] If the type of the loan product is corporate loan, determine that the loan interest rate prediction model corresponding to the loan product is the corporate loan interest rate prediction model.

[0014] Optionally, the loan interest rate prediction model is trained in the following manner:

[0015] Obtain loan interest rate correlation data;

[0016] Perform correlation analysis on the loan interest rate correlation data according to the type of the loan product to obtain loan interest rate impact variable data corresponding to different loan products;

[0017] Train the loan interest rate prediction model according to the loan interest rate impact variable data corresponding to different loan products.

[0018] Optionally, the loan interest rate prediction model includes a personal loan interest rate prediction model, and the personal loan interest rate prediction model is trained in the following manner:

[0019] According to the loan interest rate impact variable data corresponding to the personal loan and the prior knowledge data, use the Bayesian and MCMC algorithms to train the Prophet model to obtain a trained Prophet model;

[0020] Use the trained Prophet model as the personal loan interest rate prediction model.

[0021] Optionally, the loan interest rate prediction is performed using the loan interest rate prediction model corresponding to the loan product according to the loan interest rate impact variable data corresponding to the loan product, and the obtained loan interest rate prediction result includes:

[0022] If the type of the loan product is personal loan, perform loan interest rate prediction using the personal loan interest rate prediction model according to the loan interest rate impact variable data corresponding to the personal loan to obtain a personal loan interest rate prediction result;

[0023] If the type of the loan product is corporate loan, perform loan interest rate prediction using the corporate loan interest rate prediction model according to the loan interest rate impact variable data corresponding to the corporate loan to obtain a corporate loan interest rate prediction result.

[0024] Optionally, the corporate loan interest rate prediction model includes an interest rate prediction model trained based on the SARIMA model and an interest rate prediction model trained based on the LSTM model. Using the corporate loan interest rate prediction model to predict the interest rate according to the interest rate influencing variable data corresponding to the corporate loan, obtaining the corporate loan interest rate prediction result includes:

[0025] According to the interest rate influencing variable data corresponding to the corporate loan, using the interest rate prediction model trained based on the SARIMA model to predict the long-term interest rate, and obtaining the long-term interest rate prediction result;

[0026] According to the interest rate influencing variable data corresponding to the corporate loan, using the interest rate prediction model trained based on the LSTM model to predict the short-term interest rate, and obtaining the short-term interest rate prediction result.

[0027] In a second aspect, an embodiment of the present application further provides an interest rate prediction device, which includes:

[0028] An acquisition unit, configured to acquire the interest rate influencing variable data corresponding to the loan product according to the type of the loan product;

[0029] A determination unit, configured to determine the interest rate prediction model corresponding to the loan product according to the type of the loan product;

[0030] A prediction unit, configured to predict the interest rate using the interest rate prediction model corresponding to the loan product according to the interest rate influencing variable data corresponding to the loan product, and obtain the interest rate prediction result.

[0031] In a third aspect, an embodiment of the present application further provides a budget system, which includes the foregoing interest rate prediction device.

[0032] Optionally, the budget system further includes a calculation device, and the calculation device is specifically configured to:

[0033] Acquire the interest rate prediction result;

[0034] Perform calculations according to the interest rate prediction result to obtain a calculation result, and the calculations include at least one of net interest calculation, target calculation, and rolling calculation.

[0035] In a fourth aspect, an embodiment of the present application further provides a device, including:

[0036] A processor; and a memory arranged to store computer-executable instructions, and when the executable instructions are executed, the processor executes any one of the foregoing interest rate prediction methods.

[0037] In a fifth aspect, an embodiment of the present application further provides a computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the foregoing loan interest rate prediction method is implemented.

[0038] At least one of the foregoing technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: In the loan interest rate prediction method of the embodiments of the present application, first, loan product corresponding loan interest rate influence variable data is obtained according to the type of the loan product; then, a loan product corresponding loan interest rate prediction model is determined according to the type of the loan product; finally, according to the loan product corresponding loan interest rate influence variable data, the loan product corresponding loan interest rate prediction model is used to perform loan interest rate prediction to obtain a loan interest rate prediction result. The loan interest rate prediction method of the embodiments of the present application analyzes the product characteristics and data compositions of different loans, and uses different types of loan interest rate prediction models for prediction, so that both the model and the data play the greatest role, and the accuracy and reliability of loan interest rate prediction are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The drawings described herein are used to provide a further understanding of the present application, and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation to the present application. In the drawings:

[0040] Figure 1 It is a flowchart of a loan interest rate prediction method in an embodiment of the present application;

[0041] Figure 2 It is a schematic diagram of the statistical result of a dynamic default rate over time in an embodiment of the present application;

[0042] Figure 3 It is a schematic diagram of another statistical result of a dynamic default rate over time in an embodiment of the present application;

[0043] Figure 4 It is a schematic diagram of a SHAP value in an embodiment of the present application;

[0044] Figure 5 It is a schematic diagram of a comparison of model prediction results in an embodiment of the present application;

[0045] Figure 6 It is a schematic diagram of the structure of a loan interest rate prediction device in an embodiment of the present application;

[0046] Figure 7 It is a schematic diagram of the overall architecture of a budget system in an embodiment of the present application;

[0047] Figure 8 It is a schematic diagram of the structure of a device in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments of this application and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts belong to the scope of protection of this application.

[0049] The following will detail the technical solutions provided by each embodiment of this application in conjunction with the drawings.

[0050] An embodiment of this application provides a loan interest rate prediction method. As Figure 1 shown, a flowchart of a loan interest rate prediction method in an embodiment of this application is provided. The loan interest rate prediction method at least includes the following steps S110 to step S130:

[0051] Step S110, obtain the loan interest rate impact variable data corresponding to the loan product according to the type of the loan product.

[0052] When predicting the loan interest rate, it is necessary to first collect the variable data that affects the loan interest rate. Different types of loan products are often affected by different factors. Therefore, it is necessary to collect relevant loan interest rate impact variable data according to the type of the loan product (such as personal loans and corporate loans).

[0053] For example, for personal loans, the variables that affect the loan interest rate may include the borrower's credit score, income level, debt situation, loan amount, loan term, etc. For corporate loans (i.e., enterprise loans), the variables that affect the loan interest rate may include the enterprise's operating conditions, financial statements, industry risks, enterprise credit ratings, loan purposes, guarantee methods, etc. Embodiments of this application can obtain these variable data related to the loan interest rate through data interfaces, internal databases, or third-party data providers, etc., to provide input for the subsequent loan interest rate prediction model.

[0054] Step S120, determine the loan interest rate prediction model corresponding to the loan product according to the type of the loan product.

[0055] Considering that there are differences in the impact variable data structures and change characteristics of different loan products, etc., embodiments of this application have constructed different loan interest rate prediction models for different types of loan products, and select an appropriate loan interest rate prediction model according to the type of the loan product to be predicted currently, so as to give full play to the role of the interest rate impact variable data and models of different types of loan products.

[0056] For example, for personal loans, since the data is relatively simple and standardized, linear regression models, decision tree models, random forest models, or simple neural network models can be selected. These models can handle the linear relationships and simple non-linear relationships in the data related to personal loan interest rates and have relatively high computational efficiency. For corporate loans, on the other hand, since the data is more complex and diverse, more complex models such as deep neural networks, gradient boosting trees, and support vector machines may be needed, or multiple models can be combined. These models can handle complex non-linear relationships and feature interactions and are more suitable for dealing with the complexity and uncertainty in corporate loan data.

[0057] Step S130: According to the variable data of the influencing factors of the loan interest rate corresponding to the loan product, use the loan interest rate prediction model corresponding to the loan product to predict the loan interest rate and obtain the loan interest rate prediction result.

[0058] After determining the loan interest rate prediction model corresponding to the loan product to be predicted currently, the variable data of the influencing factors of the loan interest rate obtained in the above step S110 can be used as the input of this model to predict the interest rate corresponding to the loan product.

[0059] The loan interest rate prediction method of the embodiment of the present application analyzes the product characteristics and data composition of different loans and uses different types of loan interest rate prediction models for prediction, enabling both the model and the data to play the greatest role and improving the accuracy and reliability of loan interest rate prediction.

[0060] In some embodiments of the present application, the determining the loan interest rate prediction model corresponding to the loan product according to the type of the loan product includes: if the type of the loan product is a personal loan, determining the loan interest rate prediction model corresponding to the loan product as a personal loan interest rate prediction model; if the type of the loan product is a corporate loan, determining the loan interest rate prediction model corresponding to the loan product as a corporate loan interest rate prediction model.

[0061] The embodiment of the present application will adopt corresponding loan interest rate prediction models according to the type of the loan product. Here, the types of loan products are mainly divided into personal loans and corporate loans.

[0062] If the type of the current loan product is a personal loan, a personal loan interest rate prediction model needs to be used to predict the loan interest rate. The personal loan interest rate prediction model can be constructed based on factor variables such as the personal credit score, income level, debt situation, loan amount, and loan term of the borrower that are significantly related to the personal loan interest rate. These factors can usually reflect the repayment ability and credit status of the borrower and thus affect the personal loan interest rate.

[0063] If the type of the current loan product is corporate loan, a corporate loan interest rate prediction model needs to be used to predict the loan interest rate. The corporate loan interest rate prediction model can be constructed based on factor variables significantly related to the corporate loan interest rate, such as the business conditions of the enterprise, financial statements, industry risks, enterprise credit ratings, loan purposes, and guarantee methods. These factors can comprehensively reflect the repayment ability and credit risk of the enterprise, thereby affecting the corporate loan interest rate.

[0064] In specific implementation, it is necessary to pre-train the interest rate prediction models for different loan products and select the corresponding model for prediction according to the type of the loan product. These models can be constructed based on machine learning or deep learning algorithms, such as linear regression, decision tree, random forest, neural network, etc. How to specifically train the interest rate prediction model can be flexibly selected by those skilled in the art according to actual needs and will not be specifically limited herein.

[0065] In addition, it should be noted that the division of loan product types in the embodiments of the present application is only an exemplary implementation. Those skilled in the art can flexibly define and divide loan product types according to actual needs. For example, personal loans can also be divided into personal housing loans, auto consumer loans, etc., which will not be enumerated one by one herein.

[0066] By constructing interest rate prediction models for different types of loan products respectively, the characteristics and influencing factors of different types of loan products can be captured more accurately, thereby improving the accuracy of prediction. Different types of loan products such as personal loans and corporate loans have significant differences in risk characteristics, approval processes, repayment methods, etc. By constructing prediction models for different loan products respectively, the needs of different types of loan products can be better adapted, and the applicability and flexibility of the model can be improved.

[0067] In some embodiments of the present application, the interest rate prediction model is trained as follows: obtain interest rate correlation data; perform correlation analysis on the interest rate correlation data according to the type of the loan product to obtain interest rate influencing variable data corresponding to different loan products; train the interest rate prediction model according to the interest rate influencing variable data corresponding to different loan products.

[0068] When training the interest rate prediction model, it is necessary to first collect the original data related to the newly occurred loan interest rate. Considering that China's loan interest rate is affected by the dual-track system, that is, it is affected by both policy pricing and internal bank pricing, the embodiments of the present application can collect the following three major categories of original data related to the newly occurred loan interest rate based on this.

[0069] 1) External market variables: Collect more than a thousand dimensions of external market variable data, including but not limited to various core CPI (Consumer Price Index), PPI (Producer Price Index) data, and macroeconomic data. These data reflect the changes in the overall economic situation and market environment and have a potential impact on personal loan interest rates.

[0070] 2) Dynamic default rate data: Obtain the monthly dynamic default rate data of the loans that have occurred in this bank. These data reflect the repayment ability and credit status of borrowers and are important indicators for assessing the risk situation of customer groups. As Figures 2-3 shown, it provides a schematic diagram of the statistical results of the dynamic default rate over time in the embodiments of this application. Taking the housing loan as an example, Figure 2 it represents the proportion of newly issued loans defaulting in each subsequent month, Figure 3 and it represents the weighted summary of the default proportions for all months.

[0071] 3) Historical interest rate data: Collect the historical interest rate data of the loans that have occurred in this bank. These data provide information on historical interest rate levels and help the model understand the trends and patterns of interest rate changes.

[0072] Use SHAP (SHapley Additive exPlanations) values to perform a correlation analysis on the collected loan interest rate correlation data. SHAP values are a tool for explaining the prediction results of machine learning models, and they can quantify the contribution of each feature to the model's prediction results. As Figure 4 shown, it provides a schematic diagram of SHAP values in the embodiments of this application. Through SHAP value analysis, variables that have a greater impact on the loan interest rates of different loan products can be identified, and the variable data significantly related to different loan products are used as the input of the loan interest rate prediction model to train the loan interest rate prediction models for different loan products.

[0073] By collecting multi-dimensional loan interest rate correlation data and using SHAP values for correlation analysis, the model can more accurately capture the key factors affecting the loan interest rates of different loan products, which helps to improve the prediction accuracy of the model and make the prediction results closer to the actual situation. SHAP value analysis not only provides information on the contribution of variables to the model's prediction results but also can generate visual explanation graphs to help users understand the working principle and prediction results of the model, enhancing the interpretability of the model, enabling financial institutions to better understand and trust the prediction results of the model, and thus being able to flexibly adjust loan interest rates according to factors such as the credit status of borrowers and the market environment to balance risks and returns.

[0074] In some embodiments of the present application, the loan interest rate prediction model includes an individual loan interest rate prediction model, and the individual loan interest rate prediction model is trained in the following manner: According to the loan interest rate impact variable data and prior knowledge data corresponding to individual loans, use Bayesian and MCMC algorithms to train the Prophet model to obtain a trained Prophet model; use the trained Prophet model as the individual loan interest rate prediction model.

[0075] For individual loan data, because of its large amount of data, it conforms to the law of large numbers and the central limit theorem. At the same time, there are often prior knowledge in the banking industry loan that conform to the realistic laws of the banking industry, such as the good start of the year, the peak and off-peak seasons, and the pre-estimated values of the interest rate volatility of various products by industry experts. Therefore, in order to effectively utilize this prior knowledge and the cycle and seasonal effects, the individual loan interest rate prediction model in the embodiments of the present application can use the Prophet model as the basic model for training.

[0076] The Prophet model is a model developed by Facebook for time series prediction. It is particularly good at dealing with time series data with periodic changes and holiday effects. By combining Bayesian methods and Markov Chain Monte Carlo (MCMC) algorithms, prior knowledge can be more effectively incorporated into the model training process. Bayesian methods allow considering prior distributions during model training, while the MCMC algorithm is used to sample from the posterior distribution to obtain more accurate parameter estimates. During the training process, the loan interest rate impact variable data and prior knowledge data of individual loans are used as inputs, and the parameters of the Prophet model are adjusted to minimize the prediction error, thereby outputting a trained individual loan interest rate prediction model.

[0077] Of course, in addition to using the Prophet model, those skilled in the art can also flexibly select other types of time series prediction models according to actual needs, which will not be listed one by one here.

[0078] By combining prior knowledge with strong time series characteristics and loan interest rate impact variable data, the model can better capture the complexity and dynamics of the loan market, thereby improving the accuracy of individual loan interest rate prediction. The introduction of prior knowledge helps to enhance the interpretability of the model, making it easier for bank practitioners to understand and trust the prediction results of the model. The Prophet model is good at dealing with periodic changes and holiday effects in time series data, which enables it to better adapt to the seasonal fluctuations of the loan market and the impact of unexpected events (such as economic recessions, policy adjustments, etc.) on loan interest rates.

[0079] To verify the effect of the loan interest rate prediction model trained in the above embodiments, such as Figure 5As shown, a schematic diagram for comparing model prediction results in an embodiment of the present application is provided. In the embodiment of the present application, in combination with Asset Profilio analysis, customer group risk analysis and Prophet prediction model are used, and the macroeconomic impact is carefully considered. Backtesting with historical data shows that the annualized interest rate of the core product fluctuates within 10bp and the variance is controllable.

[0080] In some embodiments of the present application, the predicting the loan interest rate by using the loan interest rate prediction model corresponding to the loan product according to the loan interest rate influencing variable data corresponding to the loan product includes: if the type of the loan product is personal loan, predicting the loan interest rate by using the personal loan interest rate prediction model according to the loan interest rate influencing variable data corresponding to the personal loan to obtain the personal loan interest rate prediction result; if the type of the loan product is corporate loan, predicting the loan interest rate by using the corporate loan interest rate prediction model according to the loan interest rate influencing variable data corresponding to the corporate loan to obtain the corporate loan interest rate prediction result.

[0081] Based on the foregoing embodiments, when predicting the loan interest rate, the corresponding loan interest rate prediction model can be called according to the current loan product type for interest rate prediction. For example, if it is a personal loan, the personal loan interest rate prediction result is obtained by predicting the loan interest rate by using the personal loan interest rate prediction model according to the loan interest rate influencing variable data corresponding to the personal loan; if it is a corporate loan, the corporate loan interest rate prediction result is obtained by predicting the loan interest rate by using the corporate loan interest rate prediction model according to the loan interest rate influencing variable data corresponding to the corporate loan.

[0082] By using different loan interest rate prediction models for different types of loan products, the characteristics and influencing factors of different loan products can be captured more accurately, thereby improving the accuracy of loan interest rate prediction. Different types of loan products have different characteristics and risk factors. Using specialized prediction models can better adapt to these differences and improve the applicability and flexibility of the models.

[0083] In some embodiments of the present application, the corporate loan interest rate prediction model includes an interest rate prediction model trained based on the SARIMA model and an interest rate prediction model trained based on the LSTM model. Using the corporate loan interest rate prediction model to predict the interest rate according to the interest rate influencing variable data corresponding to the corporate loan, the corporate loan interest rate prediction result obtained includes: using the interest rate prediction model trained based on the SARIMA model to predict the long-term interest rate according to the interest rate influencing variable data corresponding to the corporate loan, and obtaining the long-term interest rate prediction result; using the interest rate prediction model trained based on the LSTM model to predict the short-term interest rate according to the interest rate influencing variable data corresponding to the corporate loan, and obtaining the short-term interest rate prediction result.

[0084] For corporate loans, due to the small amount of data and the existence of single-item pricing, etc., in terms of model selection, a combination of the SARIMA model and the LSTM model is adopted. When constructing the SARIMA model, more consideration is given to the past data trend, which can capture the seasonal trend and long-term dependence in time series data, so it is more suitable for predicting long-term loan interest rates. Since the LSTM model has more complex model parameters and is greatly affected by adjacent data, it is particularly good at dealing with short-term dependence and non-linear relationships in time series data. Therefore, while being more suitable for predicting corporate loan interest rates, it is used for accurate prediction of interest rates within a short cycle.

[0085] By combining the SARIMA model and the LSTM model, the respective advantages can be fully utilized to capture the long-term trend and short-term fluctuations respectively, thereby improving the accuracy of corporate loan interest rate prediction. Different types of corporate loans and different market environments may have different requirements for long-term and short-term predictions. By combining the two models, these changes can be more flexibly adapted, and the adaptability of the model can be improved.

[0086] The embodiments of the present application also provide a loan interest rate prediction device 600, as Figure 6 shown, which provides a structural schematic diagram of a loan interest rate prediction device in the embodiments of the present application. The loan interest rate prediction device 600 includes: an acquisition unit 610, a determination unit 620, and a prediction unit 630, where:

[0087] The acquisition unit 610 is configured to acquire the interest rate influencing variable data corresponding to the loan product according to the type of the loan product;

[0088] The determination unit 620 is configured to determine the loan interest rate prediction model corresponding to the loan product according to the type of the loan product;

[0089] A prediction unit 630, configured to predict the loan interest rate by using the loan interest rate prediction model corresponding to the loan product based on the loan interest rate influencing variable data corresponding to the loan product, so as to obtain a loan interest rate prediction result.

[0090] In some embodiments of the present application, the determining unit 620 is specifically configured to: if the type of the loan product is a personal loan, determine that the loan interest rate prediction model corresponding to the loan product is a personal loan interest rate prediction model; if the type of the loan product is a corporate loan, determine that the loan interest rate prediction model corresponding to the loan product is a corporate loan interest rate prediction model.

[0091] In some embodiments of the present application, the loan interest rate prediction model is trained in the following manner: obtaining loan interest rate correlation data; performing a correlation analysis on the loan interest rate correlation data according to the type of the loan product to obtain loan interest rate influencing variable data corresponding to different loan products; training the loan interest rate prediction model according to the loan interest rate influencing variable data corresponding to different loan products.

[0092] In some embodiments of the present application, the loan interest rate prediction model includes a personal loan interest rate prediction model, and the personal loan interest rate prediction model is trained in the following manner: training a Prophet model by using the loan interest rate influencing variable data corresponding to the personal loan and prior knowledge data by using Bayesian and MCMC algorithms to obtain a trained Prophet model; using the trained Prophet model as the personal loan interest rate prediction model.

[0093] In some embodiments of the present application, the prediction unit 630 is specifically configured to: if the type of the loan product is a personal loan, predict the loan interest rate by using the personal loan interest rate prediction model according to the loan interest rate influencing variable data corresponding to the personal loan to obtain a personal loan interest rate prediction result; if the type of the loan product is a corporate loan, predict the loan interest rate by using the corporate loan interest rate prediction model according to the loan interest rate influencing variable data corresponding to the corporate loan to obtain a corporate loan interest rate prediction result.

[0094] In some embodiments of the present application, the corporate loan interest rate prediction model includes an interest rate prediction model trained based on the SARIMA model and an interest rate prediction model trained based on the LSTM model. The prediction unit 630 is specifically configured to: based on the interest rate impact variable data corresponding to the corporate loan, use the interest rate prediction model trained based on the SARIMA model to perform long-term loan interest rate prediction to obtain a long-term loan interest rate prediction result; based on the interest rate impact variable data corresponding to the corporate loan, use the interest rate prediction model trained based on the LSTM model to perform short-term loan interest rate prediction to obtain a short-term loan interest rate prediction result.

[0095] In some embodiments of the present application, the prediction unit 630 is specifically configured to: obtain customer group risk impact variable data; based on the customer group risk impact variable data, use a maturity analysis model to perform customer group risk analysis to obtain a customer group risk analysis result; based on the interest rate impact variable data corresponding to the loan product and the customer group risk analysis result, use the interest rate prediction model corresponding to the loan product to perform loan interest rate prediction to obtain the loan interest rate prediction result.

[0096] In some embodiments of the present application, the loan interest rate prediction device 600 further includes: a calculation unit, configured to, after using the interest rate prediction model corresponding to the loan product to perform loan interest rate prediction based on the interest rate impact variable data corresponding to the loan product to obtain a loan interest rate prediction result, perform calculations based on the loan interest rate prediction result to obtain a calculation result, where the calculations include at least one of net interest calculation, target calculation, and rolling calculation.

[0097] It can be understood that the above loan interest rate prediction device can implement each step of the loan interest rate prediction method provided in the foregoing embodiments. The relevant explanations regarding the loan interest rate prediction method are applicable to the loan interest rate prediction device and will not be elaborated here.

[0098] The embodiments of the present application further provide a budget system, and the budget system includes the foregoing loan interest rate prediction device.

[0099] The loan interest rate information predicted based on the loan interest rate prediction device in the present application can be applied to net interest calculation, target calculation, and rolling calculation in the budget system, and at the same time supports independent viewing in the budget analysis module, becoming an important technical support tool for the comprehensive budget management system. The comprehensive budget system covers modules such as annual budget, flexible calculation, budget analysis, report management, and public management, aiming to provide scientific and systematic management support for the realization of the strategic goals of the entire bank.

[0100] In the measurement process of this application, macroeconomic factors are fully combined with the quantity, price, and risk characteristics of deposit and loan products. Through innovative algorithm and model design, scientific and accurate prediction of the newly generated customer interest rates of various products is achieved. It not only provides a systematic theoretical method and technical solution for budget measurement, but also significantly improves the efficiency and accuracy of budget preparation and resource allocation, providing important assistance for the optimization of key links in the comprehensive budget management system. At the same time, this application lays a solid foundation for the steady progress of the bank's overall business activities and the achievement of strategic goals, fully demonstrating its value and application prospects in the modern bank budget management system.

[0101] In some embodiments of this application, the budget system further includes a measurement device, which is specifically used for: obtaining the loan interest rate prediction result; performing measurement based on the loan interest rate prediction result to obtain a measurement result, and the measurement includes at least one of net interest measurement, target measurement, and rolling measurement.

[0102] As Figure 7 shown, the overall architecture schematic diagram of a budget system in an embodiment of this application is provided. The budget system in the embodiment of this application further includes a measurement device, and the measurement device includes but is not limited to: net interest measurement, target measurement, and rolling measurement. By predicting the newly generated customer interest rates, the prices of the bank's total stock business and newly generated business are measured, so as to participate in the income measurement work of the whole bank and branches. This process reflects the difference between the interest income obtained by the bank from its assets and the interest cost paid to depositors and other sources of funds, thus reflecting the bank's profitability level. This application helps managers predict future revenue and expenditure situations, conduct effective financial planning, and achieve reasonable allocation of resources. This application is closely related to the long-term strategic planning of the enterprise, ensuring the consistency of financial goals and overall strategic goals.

[0103] In addition, by applying it to rolling measurement, the system supports managers to monitor the budget execution progress and master the actual business situation. Grasp the future development trend of the enterprise in dynamic budgeting, and assist managers at all levels to conduct detailed consideration and overall planning for the production and operation activities in a certain future period. As time goes by, the system administrator can continuously adjust and revise the business strategy, optimize resource allocation, make the budget more adaptable to the actual situation, so as to give full play to the guiding and controlling role of the budget. The system timely provides the decision-making level with a prediction picture of future business operations, helping the enterprise better adapt to market changes and improve the accuracy and efficiency of decision-making.

[0104] In summary, the key points of this application mainly include:

[0105] 1. In terms of the model:

[0106] (1) For the interest rate prediction of bank loan products, this application does not generally adopt the same model. Instead, it combines the attributes of loan products and data characteristics, and respectively uses two major methods of time series prediction combined with Bayesian and deep learning for prediction.

[0107] (2) In terms of model tuning, prior knowledge that conforms to the realistic laws of the banking industry, such as the estimated values of the opening period, peak and off-peak seasons, and industry experts' expectations for the interest rate volatility of various products themselves, is introduced to ensure the interpretability of the model.

[0108] 2. System aspect:

[0109] This application provides a scientific reference basis for the newly occurred customer interest rate during the annual budget filling process by constructing a model and flexibly integrating multi-dimensional related factors, supports the construction of a comprehensive and scientific budget system, and provides strong technical support for comprehensive budget management. This system supports the independent operation and visual display of the model management module, enabling business personnel to view relevant results monthly according to actual needs and efficiently apply them in rolling calculations. This technology covers the entire closed-loop management process from budget formulation to execution tracking, and is an indispensable support tool in the comprehensive budget management system, providing reliable guarantees for refined budget management and scientific decision-making.

[0110] This application has achieved at least the following technical effects:

[0111] (1) For time series prediction widely used in the industry using deep learning algorithms, this application analyzes the characteristics of the data itself and adopts a method combined with Bayesian for prediction, which not only improves the time complexity and space complexity of the model during training, but also ensures the prediction accuracy of the model.

[0112] (2) The model also successfully and pioneeringly introduces knowledge such as the opening period, peak and off-peak seasons, and product characteristics themselves, which have always been recognized by banking experts but cannot be applied to the model. This not only increases the accuracy of the model prediction results, but also reduces the volatility of the model prediction results.

[0113] (3) For different types of loan products, this application analyzes the product characteristics and data composition, and combines different types of models, so that both the model and the data play the greatest role, and a relatively complete interest rate prediction system for various bank loan products is formed.

[0114] Figure 8 It is a schematic structural diagram of a device in an embodiment of this application. As Figure 8 shown, the device includes one or more processors (or processing units), and may also include one or more memories coupled to the processor, and may also include a communication module coupled to the processor.

[0115] The communication module can be used to communicate with other devices or apparatuses, such as sending or receiving data and / or signals. The communication module can have at least one communication module for communication. The communication module can include any interfaces necessary for communicating with other devices. Exemplarily, the communication module can be a transceiver, a circuit, a bus, a module, or other types of communication modules.

[0116] The processor can include, but is not limited to, at least one of the following: a general-purpose computer, a special-purpose computer, a microcontroller, a Digital Signal Processor (DSP), or one or more in a multi-core controller architecture based on a controller. The device can have multiple processors, such as an application-specific integrated circuit chip, which is subordinate to a clock synchronized with the main processor in time.

[0117] The memory can include one or more non-volatile memories and one or more volatile memories. Examples of non-volatile memories include, but are not limited to, at least one of the following: Read-Only-Memory (ROM), Electrically Programmable Read-Only-Memory (EPROM), flash memory, a hard disk, a Compact Disc (CD), a Digital Video Disk (DVD), or other magnetic storage and / or optical storage. Examples of volatile memories include, but are not limited to, at least one of the following: Random Access Memory (RAM), or other volatile memories that do not persist during a power-off duration.

[0118] The computer program includes computer-executable instructions executed by an associated processor. The program can be stored in the ROM. The processor can execute any suitable actions and processes by loading the program into the RAM.

[0119] Possible implementations of the present application can be realized by means of the program, such that the communication device can execute any process discussed in the foregoing embodiments. Possible implementations of the present application can also be realized by hardware or by a combination of software and hardware.

[0120] In some embodiments, the program can be tangibly embodied in a computer-readable storage medium, which can be included in the device (such as in the memory) or other storage devices accessible by the device. The program can be loaded from the computer-readable storage medium into the RAM for execution. The computer-readable storage medium can include any type of tangible non-volatile memory, such as ROM, EPROM, flash memory, a hard disk, a CD, a DVD, etc.

[0121] The embodiments of the present application also provide a computer-readable storage medium, on which computer instructions or program codes are stored. When the processor runs the instructions or the program codes, the processor is caused to execute the methods and functions involved in any of the above embodiments. The computer-readable medium can be any tangible medium that contains or stores a program for or related to an instruction execution system, apparatus, or device. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any suitable combination thereof. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that incorporates one or more available media. More specific examples of the computer-readable storage medium include electrical connections with one or more wires, magnetic media (such as disks, floppy disks, hard disks, magnetic tapes, magnetic storage devices), optical media (such as optical storage devices, DVDs), semiconductor media (such as solid-state drives), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), or any suitable combination thereof, etc.

[0122] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The embodiments of the present application also provide at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes one or more computer-executable instructions, such as the instructions included in the program module, which are executed in a device on a target real or virtual processor to execute the processes, methods, and functions involved in any of the above embodiments. When the computer program instructions are loaded and executed on a computer, the processes or functions according to the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line) or wirelessly (such as infrared, wireless, microwave, etc.).

[0123] Embodiments of the present application also propose a computer program product, including a computer program or instructions. When the computer program or instructions run on a computer, the computer is caused to execute the processes, methods, and functions in the above embodiments. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules can be combined or divided as needed. Machine-executable instructions for program modules can be executed within local or distributed devices. In a distributed device, program modules can be located in local and remote storage media.

[0124] Generally, various embodiments of the present application can be implemented in hardware or special-purpose circuits, software, logic, or any combination thereof. Some aspects can be implemented in hardware, while other aspects can be implemented in firmware or software, which can be executed by a controller, a microprocessor, or other computing devices. Although aspects of the embodiments of the present disclosure are shown and described as block diagrams, flowcharts, or using some other graphical representation, it should be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as, by way of non-limiting example, hardware, software, firmware, special-purpose circuits or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0125] It should be noted that although the embodiments of the present application are described above in conjunction with the accompanying drawings respectively, the above embodiments are not independent of each other, and they can also be combined to obtain other embodiments. The manners, situations, categories, and divisions of the embodiments in the present application are only for the convenience of description and should not constitute a special limitation. The features in various manners, categories, situations, and embodiments can be combined with each other under logical conditions. The various embodiments of the present application can be combined arbitrarily to achieve different technical effects. The embodiments of the present application will no longer list various combinations.

[0126] In addition, although the operations of the methods of the present disclosure are described in a specific order in the drawings, this does not require or imply that these operations must be performed in that specific order, or that all the illustrated operations must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart can change the execution order. Additionally or alternatively, some steps can be omitted, multiple steps can be combined into one step for execution, and / or one step can be decomposed into multiple steps for execution. It should also be noted that the features and functions of two or more apparatuses according to the present disclosure can be embodied in one apparatus. Conversely, the features and functions of one apparatus described above can be further divided and embodied by multiple apparatuses.

[0127] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising said element.

[0128] The above description is only for the embodiments of the present application and is not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A loan interest rate prediction method, characterized in that: The loan interest rate prediction method comprises: Obtaining loan interest rate influencing variable data corresponding to the loan product according to the type of the loan product; Determining a loan interest rate prediction model corresponding to the loan product according to the type of the loan product; According to the loan interest rate influencing variable data corresponding to the loan product, the loan interest rate prediction model corresponding to the loan product is used to perform loan interest rate prediction to obtain a loan interest rate prediction result.

2. The loan interest rate prediction method according to claim 1, characterized in that: Determining the loan interest rate prediction model corresponding to the loan product according to the type of the loan product includes: If the type of the loan product is a personal loan, determining that the loan interest rate prediction model corresponding to the loan product is a personal loan interest rate prediction model; If the type of the loan product is a corporate loan, then the loan interest rate prediction model corresponding to the loan product is determined to be a corporate loan interest rate prediction model.

3. The loan interest rate prediction method according to claim 1, characterized in that: The loan interest rate prediction model is trained in the following way: Get loan interest rate related data; Performing correlation analysis on the loan interest rate related data according to the type of loan products to obtain loan interest rate influencing variable data corresponding to different loan products; The loan interest rate prediction model is trained according to the loan interest rate influencing variable data corresponding to different loan products.

4. The loan interest rate prediction method according to claim 3, characterized in that: The loan interest rate prediction model includes a personal loan interest rate prediction model, and the personal loan interest rate prediction model is trained in the following manner: According to the variable data and prior knowledge data of the loan interest rate influencing the personal loan, the Prophet model is trained using the Bayesian and MCMC algorithms to obtain a trained Prophet model; The trained Prophet model is used as the personal loan interest rate prediction model.

5. The loan interest rate prediction method according to claim 1, characterized in that: The loan interest rate prediction result obtained by performing loan interest rate prediction based on the loan interest rate influencing variable data corresponding to the loan product and using the loan interest rate prediction model corresponding to the loan product includes: If the type of the loan product is a personal loan, then according to the loan interest rate influencing variable data corresponding to the personal loan, a personal loan interest rate prediction model is used to predict the loan interest rate to obtain a personal loan interest rate prediction result; If the type of the loan product is a corporate loan, the loan interest rate is predicted using a corporate loan interest rate prediction model based on the loan interest rate influencing variable data corresponding to the corporate loan to obtain a corporate loan interest rate prediction result.

6. The loan interest rate prediction method according to claim 5, characterized in that: The public loan interest rate prediction model includes a loan interest rate prediction model obtained by training based on the SARIMA model and a loan interest rate prediction model obtained by training based on the LSTM model. The public loan interest rate prediction model is used to predict the loan interest rate according to the loan interest rate influencing variable data corresponding to the public loan, and the public loan interest rate prediction result obtained includes: According to the loan interest rate influencing variable data corresponding to the corporate loan, a loan interest rate prediction model obtained by training based on the SARIMA model is used to predict the long-term loan interest rate to obtain a long-term loan interest rate prediction result; According to the loan interest rate influencing variable data corresponding to the corporate loan, a loan interest rate prediction model obtained by training based on the LSTM model is used to predict the short-term loan interest rate to obtain a short-term loan interest rate prediction result.

7. A loan interest rate prediction device, characterized in that: The loan interest rate prediction device comprises: An acquisition unit, used for acquiring the loan interest rate influencing variable data corresponding to the loan product according to the type of the loan product; A determination unit, configured to determine a loan interest rate prediction model corresponding to the loan product according to the type of the loan product; The prediction unit is used to predict the loan interest rate according to the loan interest rate influencing variable data corresponding to the loan product and use the loan interest rate prediction model corresponding to the loan product to obtain the loan interest rate prediction result.

8. A budget system, characterized in that: The budget system includes the loan interest rate prediction device according to claim 7.

9. The budget system according to claim 8, characterized in that: The budget system further includes a calculation device, which is specifically used for: Get loan interest rate prediction results; A calculation result is obtained based on the loan interest rate forecast result, and the calculation includes at least one of net interest calculation, target calculation and rolling calculation.

10. A device comprising: processor; And a memory arranged to store computer executable instructions, which, when executed, cause the processor to execute any one of the loan interest rate prediction methods of claims 1 to 6.

11. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the loan interest rate prediction method according to any one of claims 1 to 6 is implemented.

Images