Industry and finance integrated data management system and method based on big data analysis

Abstract

The application discloses a business and finance integrated data management system and method based on big data analysis, and belongs to the technical field of enterprise resource management. The system comprises: a quotation access module, which is used for receiving a quotation request carrying a customer identifier and quotation content in real time; a business and finance data fusion module, which generates a business and finance fusion feature vector; a customer classification module, which is used for identifying a customer type according to the business and finance fusion feature vector; a system load perception module, which calculates a predicted waiting time of the current quotation; a customer behavior prediction module, which calculates a quotation survival probability according to the predicted waiting time; a value quantification and risk adjustment module, which obtains an opportunity value after risk adjustment; a priority comprehensive calculation module, which calculates a comprehensive priority index of the quotation; and a strategy scheduling module, which matches a differentiated response strategy according to the comprehensive priority index and the customer type difference. The application can effectively balance the transaction conversion rate improvement and financial risk control, and improve the overall resource utilization efficiency and income level of enterprises.

Description

Technical Field

[0001] This invention relates to the field of enterprise resource management technology, specifically to an integrated business and finance data management system and method based on big data analysis. Background Technology

[0002] In daily operations, businesses often face a large number of price inquiries from different customers. Traditional inquiry processing methods often rely on manual judgment or first-come-first-served queue rules, failing to systematically prioritize inquiry requests or deeply integrate business data with financial data. This leads to the following problems: high-value or high-credit customers may be lost due to untimely responses; inquiries from customers with operational risks or low credit consume significant processing resources but yield low actual returns; and some customers abandon their inquiries due to waiting times exceeding their psychological tolerance, further reducing conversion rates.

[0003] Existing technologies include fixed priority settings based on customer classification. However, these solutions fail to dynamically incorporate the current system load, changes in customer behavior during the waiting process, and the adjusted returns from the financial risks associated with the inquiry. Consequently, the priority calculation results are not accurate enough, making it difficult to maximize the overall benefits for the enterprise.

[0004] Therefore, those skilled in the art have provided an integrated business and finance data management system and method based on big data analysis to solve the problems mentioned in the background art. Summary of the Invention

[0005] The purpose of this invention is to provide an integrated business and financial data management system and method based on big data analysis, in order to solve the problems of priority calculation in the prior art that fails to integrate business and financial data, and does not consider customer waiting and abandonment behavior and system load status, resulting in priority distortion and resource mismatch.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The business and finance integrated data management system based on big data analytics includes: The inquiry access module is used to receive inquiry requests carrying customer identifiers and inquiry content in real time; The business and financial data fusion module is used to extract business transaction records and financial credit records from the business database and financial database respectively based on the customer identifier, align the business data and financial data according to the timestamp, so that the business records and financial records under the same customer identifier correspond to the same time window, and extract preset feature fields from the aligned data to generate a business and financial fusion feature vector. The customer classification module is used to identify customer types based on the business-finance fusion feature vector, and the customer types include new customers, old customers, healthy business customers, risky business customers, high-credit customers and low-credit customers. The system load awareness module is used to monitor the number of pending inquiries and the average processing rate in the inquiry processing queue in real time, and to calculate the predicted waiting time for the current inquiry. The customer behavior prediction module is used to predict the customer's inquiry abandonment risk function under the current system state by using the business and finance fusion feature vector and the pre-trained survival analysis model, and to calculate the inquiry survival probability based on the predicted waiting time. The value quantification and risk adjustment module is used to predict the benchmark transaction probability and expected gross profit of the inquiry based on the business-finance integration feature vector, the inquiry content and the inquiry survival probability, calculate the expected credit loss combined with financial credit, and obtain the risk-adjusted opportunity value accordingly. The priority calculation module is used to calculate the comprehensive priority index of the inquiry by using the risk-adjusted opportunity value, combined with the timeliness decay factor generated based on the predicted waiting time and the estimated service resource consumption. The strategy scheduling module is used to match differentiated response strategies based on the comprehensive priority index and the customer type, and to allocate inquiries to processing terminals in descending order of the comprehensive priority index.

[0007] As a further aspect of the present invention, the business and financial data fusion module further includes: extracting transaction frequency, average transaction amount, return rate, and gross profit contribution from the business side within the most recent preset period, and accounts receivable turnover days, overdue number, aging structure, and external credit rating score from the financial side; filling missing values ​​and standardizing continuous fields, and calculating a weighted average of time series data with greater weight the closer to the current time, wherein the weight decays exponentially with historical time intervals, and outputting a business and financial fusion feature vector containing a business vitality index and a financial stability index.

[0008] As a further aspect of the present invention: the customer behavior prediction module includes: The model building unit collects customer waiting time, abandonment status, and transaction outcomes from historical inquiry records. Using the business-finance integration feature vector and waiting time as covariates, a Cox proportional hazards model is constructed. The parameters of the Cox proportional hazards model are solved using partial likelihood estimation to obtain the benchmark risk function. and covariate coefficient vector ,in For a baseline instantaneous abandonment risk rate that changes only over time and is independent of covariates, This is a vector consisting of the regression coefficients corresponding to each covariate; The abandonment risk calculation unit is used to, for the current inquiry, substitute its business-finance integration feature vector X into the Cox proportional risk model to obtain the instantaneous abandonment risk rate function. ,in for The inner product of the transpose of X and X; The survival probability calculation unit is used to obtain the predicted waiting time. By analyzing the instantaneous abandonment risk rate function from zero to... Integrate over the interval and take the exponent to calculate the customer's waiting time. The probability of survival if one does not give up within a certain time. The calculation process is expressed as follows: ; Where u is the integration variable, representing time.

[0009] As a further aspect of the present invention: the benchmark transaction probability in the value quantification and risk adjustment module It is obtained through a pre-trained gradient boosting tree model, which takes the business-finance fusion feature vector and the product type being inquired about as input features and outputs a predicted transaction probability value; the value quantification and risk adjustment module also calculates the dynamic transaction probability, the dynamic transaction probability... The product of the baseline transaction probability and the inquiry survival probability is expressed as follows: .

[0010] As a further aspect of the present invention: the process by which the value quantification and risk adjustment module obtains the risk-adjusted opportunity value includes: Based on the estimated transaction amount A from the inquiry, and using the historical gross profit margin in the business-finance integration feature vector and the current market reference index to predict the current gross profit margin m, where the current market reference index is an external data indicator reflecting industry supply and demand and price fluctuations, the nominal opportunity profit is calculated. ; Using the overdue status transition matrix constructed from the overdue transfer ratios of different aging periods in the financial data, and combined with a pre-trained logistic regression model, the probability of default PD for this transaction is predicted. Combining the risk exposure EAD=A×(1-prepayment ratio) and the preset default loss rate LGD, the expected credit loss ECL=EAD×PD×LGD is calculated, where the prepayment ratio is the proportion of the prepayment required before the transaction to the transaction amount, which is preset according to the customer type. The risk-adjusted opportunity value is obtained by deducting the expected credit loss from the nominal opportunity profit. .

[0011] As a further aspect of the present invention: the system load perception module determines a priority adjustment factor θ based on the number of pending inquiries Q in the current queue and the average processing rate μ, and according to the customer type output by the customer classification module. Specifically, θ for high-credit customers and healthy business customers is set to a first value, and θ for low-credit customers and business-risk customers is set to a second value greater than the first value, through a relational formula. Calculate the predicted waiting time.

[0012] As a further aspect of the present invention: the priority comprehensive calculation module includes: The time loss calculation unit is used to calculate the predicted waiting time. Compared to the preset average acceptable response time for customers based on industry statistics Comparison, generating the aging decay factor γ, the calculation process is as follows: when Greater than hour, Otherwise γ=1, where This is the decay rate control parameter; The composite index calculation unit is used to calculate the risk-adjusted opportunity value. Multiplying the result by the time-depreciation factor γ and then dividing by the sum of the estimated service resource consumption C and the zero-prevention constant ε, we obtain the comprehensive priority index P. The calculation process is expressed as follows: ; The estimated service resource consumption C is obtained by mapping the preset complexity level of the materials in the inquiry content and the average time taken to process the customer's historical inquiries, and ε is a preset positive number.

[0013] As a further aspect of the present invention: the comprehensive index calculation unit also dynamically adjusts the attenuation rate control parameters according to the customer type. When a customer is identified as a high-credit customer or a customer with healthy business practices, reduce... The value of is increased when a customer is identified as a high-risk or low-credit customer. The value of .

[0014] As a further aspect of the present invention: the strategy scheduling module includes a preemptive scheduling submodule, which is used to pause the processing of the lowest index query when the comprehensive priority index of a new query exceeds the lowest index among all queries being processed and exceeds a preset ratio, retain its processed progress status, update its predicted waiting time to the sum of the waiting time and the waiting time re-estimated by the system load perception module using the current queue status, recalculate the survival probability and comprehensive priority index, and then insert it back into the queue.

[0015] This application also discloses a business-finance integrated data management method based on big data analytics, applied to a business-finance integrated data management system based on big data analytics, including the following steps: Receive inquiry requests in real time, carrying customer identification and inquiry details; Based on the customer identifier, business transaction records and financial credit records are extracted from the business database and financial database respectively. The business data and financial data are aligned by timestamp so that the business records and financial records under the same customer identifier correspond to the same time window. Preset feature fields are extracted from the aligned data to generate a business-finance fusion feature vector. Customer types are identified based on the business-finance integration feature vector, which includes new customers, old customers, healthy business customers, risky business customers, high-credit customers, and low-credit customers. Real-time monitoring of the number of pending inquiries and average processing rate in the inquiry processing queue; calculation of the predicted waiting time for the current inquiry. Using the business-finance fusion feature vector and the pre-trained survival analysis model, the inquiry abandonment risk function of customers in the current system state is predicted, and the inquiry survival probability is calculated based on the predicted waiting time. Based on the business-finance integration feature vector, the inquiry content, and the inquiry survival probability, the benchmark transaction probability and expected gross profit of the inquiry are predicted, the expected credit loss combined with financial credit is calculated, and the risk-adjusted opportunity value is obtained accordingly. Based on the predicted waiting time, a time-depletion factor is generated. The risk-adjusted opportunity value is then adjusted using the time-depletion factor and divided by the estimated service resource consumption to obtain the comprehensive priority index. Differentiated response strategies are matched based on the comprehensive priority index and the customer type, and inquiries are allocated to processing terminals in descending order of comprehensive priority index.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This application constructs a business-finance integrated feature vector, including a business vitality index and a financial robustness index, by aligning business transaction data with financial credit data over time windows and fusing features, thereby achieving precise customer classification. It uses a survival analysis model to predict the probability of customer abandonment during the waiting process and dynamically adjusts the transaction probability based on the system's real-time load status. When calculating the inquiry value, it considers not only nominal opportunity profit but also introduces expected credit loss based on the overdue state transition matrix and default probability to obtain a risk-adjusted opportunity value. Furthermore, it introduces a timeliness decay factor and estimated service resource consumption to generate a comprehensive priority index, which is then used to match differentiated response strategies. This system and method can dynamically and accurately quantify the comprehensive processing value of each inquiry, prioritizing the response time of high-value, high-credit customers with limited resources, while simultaneously implementing risk control strategies for low-credit customers. This effectively balances improving conversion rates with managing financial risks, thereby enhancing the overall resource utilization efficiency and profitability of the enterprise. Attached Figure Description

[0017] Figure 1 This is a structural diagram of an integrated business and finance data management system based on big data analytics. Figure 2 This is a flowchart of a business-finance integrated data management method based on big data analytics. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] As mentioned in the background section of this application, research has found that although there are existing technical solutions for setting fixed priorities based on customer classification, such solutions fail to dynamically combine the current system load status, changes in customer behavior during the waiting process, and the adjusted returns after considering the financial risks brought by the inquiry. Therefore, the priority calculation results are not accurate enough, making it difficult to maximize the overall benefits of the enterprise, and thus have certain defects.

[0020] To address the aforementioned shortcomings, this application discloses an integrated business and finance data management system and method based on big data analysis. This system can dynamically and accurately quantify the comprehensive processing value of each inquiry, prioritize response time for high-value, high-credit customers with limited resources, and simultaneously implement risk control strategies for low-credit customers. This effectively balances improving conversion rates with managing financial risks, thereby enhancing the overall resource utilization efficiency and profitability of the enterprise.

[0021] The following will describe in detail, with reference to the accompanying drawings, how the solution of this application solves the above-mentioned technical problems.

[0022] Please see Figure 1 In this embodiment of the invention, the business-finance integrated data management system based on big data analysis includes: an inquiry access module, used to receive inquiry requests carrying customer identifiers and inquiry content in real time; a business-finance data fusion module, used to extract business transaction records and financial credit records from the business database and financial database respectively based on the customer identifier, align the business data and financial data according to timestamps, so that business records and financial records under the same customer identifier correspond to the same time window, and extract preset feature fields from the aligned data to generate a business-finance fusion feature vector; a customer classification module, used to identify customer types according to the business-finance fusion feature vector, including new customers, old customers, healthy business customers, risky business customers, high-credit customers, and low-credit customers; and a system load perception module, used to monitor the number of pending inquiries and the average processing rate in the inquiry processing queue in real time, and calculate the current... The system comprises the following modules: a predicted waiting time for inquiries; a customer behavior prediction module, which uses a business-finance fusion feature vector and a pre-trained survival analysis model to predict the customer's inquiry abandonment risk function under the current system state, and calculates the inquiry survival probability based on the predicted waiting time; a value quantification and risk adjustment module, which predicts the baseline transaction probability and expected gross profit of inquiries based on the business-finance fusion feature vector, inquiry content, and inquiry survival probability, calculates the expected credit loss combined with financial credit, and obtains the risk-adjusted opportunity value accordingly; a priority comprehensive calculation module, which uses the risk-adjusted opportunity value, combined with the timeliness decay factor generated based on the predicted waiting time and the estimated service resource consumption, to calculate the comprehensive priority index of inquiries; and a strategy scheduling module, which matches differentiated response strategies according to the comprehensive priority index and customer type, and allocates inquiries to processing terminals in descending order of comprehensive priority index. This application deeply integrates business data and financial data to construct customer profiles, simulates customer waiting and abandonment behavior through a survival analysis model, calculates the survival probability of inquiries by combining the predicted waiting time under system load, and then corrects the transaction probability. It also deducts expected credit loss from the nominal opportunity profit to obtain the risk-adjusted opportunity value, and finally calculates a comprehensive priority index by comprehensively considering timeliness decay and service resource consumption. This enables dynamic and accurate sorting of inquiry requests, allowing enterprises to prioritize inquiries with higher overall benefits when resources are limited.

[0023] In this embodiment, the business and financial data fusion module further includes: extracting transaction frequency, average transaction amount, return rate, and gross profit contribution from the business side within the most recent preset period, and accounts receivable turnover days, overdue number, aging structure, and external credit rating score from the financial side; performing missing value imputation and standardization on continuous fields, and calculating a weighted average of time series data with higher weights for data closer to the current time, with the weights decreasing exponentially with historical time intervals, outputting a business and financial fusion feature vector containing a business vitality index and a financial stability index. By extracting multi-dimensional data from both the business and financial sides and performing missing value imputation, standardization, and exponentially decaying weighted averaging, business vitality and financial stability can be integrated, highlighting the impact of recent behavior and avoiding interference from outdated data. This makes the generated business and financial fusion feature vector more accurately reflect the current state of the customer, providing a reliable basis for subsequent classification and priority calculation.

[0024] Specific examples: Suppose a customer has had frequent transactions and no returns in the last 3 months of the past 12 months, but had one overdue payment 9 months ago. Define the attenuation coefficient λ = 0.1, and denot the time interval between historical data and the present as λ. (Unit: Month), the weighting calculation formula is as follows: Recent Months The weight is 1 month ago The weight is 9 months ago The weight is After Z-score standardization of fields such as transaction frequency and number of overdue payments at each time point, a weighted average is calculated according to the aforementioned weights to obtain the Business Vitality Index and Financial Stability Index. Recent high transaction frequency receives a high weight, while the weight of long-term overdue records is significantly reduced. This decaying weighting of the index means that the Financial Stability Index is mainly determined by recent payment performance, more closely reflecting the customer's current credit status.

[0025] In this embodiment, the customer behavior prediction module includes: a model building unit, used to collect customer waiting time, whether they abandoned the transaction or completed it from historical inquiry records; to construct a Cox proportional hazards model using the business-finance integration feature vector and waiting time as covariates; and to solve for the parameters of the Cox proportional hazards model using the partial likelihood estimation method to obtain the benchmark risk function. and covariate coefficient vector ,in For a baseline instantaneous abandonment risk rate that changes only over time and is independent of covariates, The vector consists of the regression coefficients corresponding to each covariate; the abandonment risk calculation unit is used to substitute the business-finance integration feature vector X into the Cox proportional risk model for the current inquiry to obtain the instantaneous abandonment risk rate function. ,in for The inner product of the transpose of X and X; the survival probability calculation unit, used to obtain the predicted waiting time. By analyzing the instantaneous abandonment risk rate function from zero to... Integrate over the interval and take the exponent to calculate the customer's waiting time. The probability of survival if one does not give up within a certain time. The calculation process is expressed as follows: Where u is the integral variable, representing time. The Cox proportional hazards model is used to predict customer inquiry abandonment behavior, dynamically quantifying the risk of customer churn due to waiting and outputting the survival probability. This provides a behavioral predictor factor that considers customer tolerance for subsequent priority calculations, avoiding misjudgments caused by relying solely on static customer classification.

[0026] Specific examples: When constructing the Cox proportional hazards model, historical inquiry data is collected. Each record includes a customer's business and financial integration feature vector X, actual waiting time, and outcome. The baseline risk function is obtained through partial likelihood estimation. and covariate coefficient vector For new inquiries, customer characteristics X = [Business Vitality Index = 1.2, Financial Stability Index = 0.8, ... ] are substituted into the Cox proportional hazards model to obtain the instantaneous abandonment risk rate function. Assuming baseline risk In the short term, it is approximately a constant value of 0.05. The calculation result is 0.6, then If the predicted waiting time output by the system load sensing module... The survival probability is calculated by integration after a few minutes: ; This indicates that there is still approximately a 63.4% probability that the customer will not give up on inquiring about the price after waiting for 5 minutes; if Extending the waiting time to 10 minutes reduced the survival rate to approximately 0.402, accurately reflecting the erosion of customer retention intentions by waiting time.

[0027] In this embodiment, the benchmark transaction probability in the value quantification and risk adjustment module It is obtained through a pre-trained gradient boosting tree model, which takes the business and finance fusion feature vector and the product type inquired as input features and outputs a predicted transaction probability value; the value quantification and risk adjustment module also calculates the dynamic transaction probability. The product of the baseline transaction probability and the survival probability of the inquiry is expressed as follows: The dynamic transaction probability is obtained by multiplying the baseline transaction probability predicted by the gradient boosting tree model with the survival probability obtained from survival analysis. This comprehensively considers the customer's own transaction tendency and the risk of abandonment during the waiting process, making the transaction probability prediction more in line with the real-time situation and improving the authenticity of subsequent value calculation.

[0028] Specific examples: A customer inquires about a certain product. The gradient boosting tree model takes the customer's feature vector and the product type as input and outputs a baseline probability of closing the deal. The customer's survival probability under the predicted waiting time is calculated as described above. The dynamic transaction probability is: ; When the system load increases, causing the predicted waiting time to increase and the survival probability to drop to 0.5, the dynamic transaction probability becomes 0.75 × 0.5 = 0.375. This quantifies the damage to transaction opportunities caused by queuing delays in real time and guides the system to prioritize processing customers who are sensitive to waiting.

[0029] In this embodiment, the process by which the value quantification and risk adjustment module obtains the risk-adjusted opportunity value includes: estimating the transaction amount A based on the inquiry content; predicting the current gross profit margin m based on the historical gross profit margin in the business-finance integration feature vector and the current market reference index (the current market reference index is an external data indicator reflecting industry supply and demand and price fluctuations); and calculating the nominal opportunity profit. Using an overdue status transition matrix constructed from the overdue transfer ratios across different aging periods in financial data, and combined with a pre-trained logistic regression model, the probability of default (PD) for this transaction is predicted. The expected credit loss (ECL) is calculated by combining the risk exposure (EAD) = A × (1 - prepayment ratio) and the preset loss on default (LGD), where the prepayment ratio is the percentage of the prepayment due before the transaction, preset according to customer type. The risk-adjusted opportunity value is obtained by deducting the expected credit loss from the nominal opportunity profit. When calculating opportunity value, deducting expected credit losses based on customer financial data from nominal opportunity profit quantifies accounts receivable recovery risk into costs, making opportunity value more risk-adjusted. This can guide companies to prioritize high-yield, low-risk inquiries and effectively control bad debt losses.

[0030] Specific examples: Customer A is a long-term, high-credit customer. The estimated transaction amount is A = 1 million yuan, the predicted gross profit margin is m = 20%, and the dynamic probability of a successful transaction is... Nominal opportunity profit The amount is 10,000 yuan. The customer's prepayment ratio is 0.3%, and the risk exposure is [missing information]. The risk-adjusted opportunity value is RMB 700,000. Based on the overdue state transition matrix and logistic regression model, the predicted default probability is PD=0.02, the loss margin of default (LGD) is 0.5, and the expected credit loss (ECL) is 70 × 0.02 × 0.5 = RMB 70,000. Ten thousand yuan.

[0031] Customer B is a low-credit customer. For the same transaction amount and gross profit margin, but with a prepayment ratio of 0.5 and a projected PD of 0.15, then EAD = 500,000 yuan, ECL = 500,000 × 0.15 × 0.5 = 37,500 yuan. Ten thousand yuan. Although the nominal profit is the same, the value of low-credit customers is reduced by about 23% due to their high credit risk, and they will be at a disadvantage in subsequent priority calculations.

[0032] In this embodiment, the system load perception module determines the priority adjustment factor θ based on the number of pending inquiries Q in the current queue and the average processing rate μ, and according to the customer type output by the customer classification module. θ for high-credit customers and healthy business customers is set to a first value, while θ for low-credit customers and business-risk customers is set to a second value greater than the first value. This is achieved through a relational formula. Calculate the predicted waiting time. The predicted waiting time is calculated by adjusting the priority factor θ based on customer type, resulting in shorter estimated waiting times for high-value customers. This gives them a relative advantage in subsequent survival probability and time-lapse calculations, reflecting the system's resource allocation towards high-value customers and improving customer experience and conversion rates.

[0033] Specific examples: The current queue has 12 pending inquiries (Q=12), and the system's average processing rate is μ=3 inquiries / minute. For high-credit customers, the adjustment factor is set to θ=0.8, and the predicted waiting time is: minute; For low-credit customers, the adjustment factor θ = 1.5, and the predicted waiting time is: minute; This difference results in high-credit customers having a lower risk of abandonment (higher survival probability) and less time-related penalties in subsequent calculations, and even in the same queue, their overall priority is more likely to be ranked higher.

[0034] In this embodiment, the priority comprehensive calculation module includes: a timeliness loss calculation unit, used to calculate the predicted waiting time. Compared to the preset average acceptable response time for customers based on industry statistics Comparison, generating the aging decay factor γ, the calculation process is as follows: when Greater than hour, Otherwise γ=1, where The decay rate control parameter; the composite index calculation unit, used to calculate the risk-adjusted opportunity value. Multiplying the result by the time-depreciation factor γ and then dividing by the sum of the estimated service resource consumption C and the zero-prevention constant ε, we obtain the comprehensive priority index P. The calculation process is expressed as follows: The estimated service resource consumption C is derived by mapping the preset complexity level of the materials in the inquiry content to the average processing time of the customer's historical inquiries, with ε being a preset positive number. By introducing a timeliness decay factor, when the predicted waiting time exceeds the customer's average tolerance time, the opportunity value is exponentially decayed, quantifying the value loss caused by response delays. Simultaneously, dividing by the service resource consumption lowers the priority of inquiries with high processing costs, achieving a comprehensive trade-off between profit and cost, and guiding precise resource allocation.

[0035] Specific examples: Preset industry average tolerance time Minutes, decay rate control parameter κ=0.5. Opportunity value after risk adjustment for a certain inquiry. The estimated service resource consumption is C=2 units, with a zero constant ε=0.001. Scenario 1: Predicted waiting time. Minutes, not exceeding the tolerance period, time decay factor γ=1, comprehensive priority index: ; Scenario 2: If the time exceeds the tolerance limit in minutes, calculate: ; ; Waiting timeouts result in an opportunity value decay of approximately 25.9%, leading to a corresponding decrease in priority. The incentive system will respond as early as possible when resources allow to avoid further value loss.

[0036] In this embodiment, the comprehensive index calculation unit also dynamically adjusts the attenuation rate control parameters according to the customer type. When a customer is identified as a high-credit customer or a customer with healthy business practices, reduce... The value of is increased when a customer is identified as a high-risk or low-credit customer. The value is determined by dynamically adjusting the attenuation rate control parameter based on the customer type. This allows the priority of high-quality customers to decay more slowly over time, preserving a longer priority response window, while the priority of high-risk customers decays more rapidly, preventing them from occupying queuing resources for a long time and further strengthening the differentiated service strategy.

[0037] Specific examples: Continuing with the previous scenario, when minutes exceeded For high-credit customers, the dynamically adjusted κ=0.2, and the time-related decay factor is: ; Priority Index For low-credit customers, κ=0.8, time-depreciation factor: ; Priority Index In the same timeout scenario, high-credit customers have a priority of approximately 43% higher than low-credit customers, and the system automatically allocates response resources to high-quality customers.

[0038] In this embodiment, the strategy scheduling module includes a preemptive scheduling submodule. When the overall priority index of a new query exceeds the lowest index among all currently processed queries and exceeds a preset proportion, the processing of the query with the lowest index is paused, its processing progress is preserved, its predicted waiting time is updated to the sum of the already waited time and the waiting time re-estimated based on the current queue status by the system load perception module, and the survival probability and overall priority index are recalculated before being reinserted into the queue. This preemptive scheduling mechanism allows extremely high-priority new queries to interrupt currently processed low-value tasks, ensuring that urgent and important queries are processed immediately, while preserving the progress of interrupted tasks and the re-estimated waiting time, balancing fairness and efficiency to maximize overall throughput benefits.

[0039] Specific example: There are currently 3 inquiries being processed, with priority indices P1=2.5, P2=3.0, and P3=1.8 (lowest). A newly arrived inquiry from a high-credit, existing customer has a calculated priority index of P... new =12.0, exceeding the minimum index of 1.8 and exceeding the preset ratio (e.g., if the ratio threshold is 1.5, 12.0 > 1.8 × 1.5 = 2.7). Preemption mechanism triggered: Pause the P3 task, record its processed time as 2 minutes; update the predicted waiting time of the task to the already waited 2 minutes plus the waiting time re-estimated based on the current queue (assuming the new estimate is 5 minutes, total waiting time...). (Minutes) Recalculate the survival probability (assuming it decreases from 0.9 to 0.6) and priority index (assuming it decreases to 1.2), and then insert it back into the queue. Immediately release resources to process new inquiries. After completion, schedule subsequent tasks from the queue according to the updated priority order to achieve rapid response to high-value inquiries and maximize the overall system efficiency.

[0040] like Figure 2As shown, this application also discloses a business-finance integrated data management method based on big data analysis, applied to a business-finance integrated data management system based on big data analysis, including the following steps: receiving inquiry requests carrying customer identifiers and inquiry content in real time; extracting business transaction records and financial credit records from the business database and financial database respectively based on the customer identifier, aligning the business data and financial data by timestamp, so that business records and financial records under the same customer identifier correspond to the same time window, and extracting preset feature fields from the aligned data to generate a business-finance integrated feature vector; identifying customer types based on the business-finance integrated feature vector, including new customers, old customers, healthy business customers, risky business customers, high-credit customers, and low-credit customers; and monitoring the pending inquiries in the inquiry processing queue in real time. Based on the number of inquiries and average processing rate, the predicted waiting time for the current inquiry is calculated. Using the business-finance fusion feature vector and a pre-trained survival analysis model, the inquiry abandonment risk function of customers in the current system state is predicted, and the inquiry survival probability is calculated based on the predicted waiting time. Based on the business-finance fusion feature vector, inquiry content, and inquiry survival probability, the baseline transaction probability and expected gross profit of the inquiry are predicted, and the expected credit loss combined with financial credit is calculated, thereby obtaining the risk-adjusted opportunity value. A timeliness decay factor is generated based on the predicted waiting time. The risk-adjusted opportunity value, after being adjusted by the timeliness decay factor, is divided by the estimated service resource consumption to obtain the comprehensive priority index. Differentiated response strategies are matched according to the comprehensive priority index and customer type, and inquiries are allocated to processing terminals in descending order of comprehensive priority index.

[0041] This invention aligns business transaction data with financial credit data through time windows and feature fusion to construct a business-finance integrated feature vector containing a business vitality index and a financial stability index, enabling precise customer classification. It uses a survival analysis model to predict the probability of customer abandonment during the waiting process and dynamically adjusts the transaction probability based on the system's real-time load status. When calculating the value of an inquiry, it considers not only nominal opportunity profit but also introduces expected credit loss based on the overdue state transition matrix and default probability to obtain a risk-adjusted opportunity value. Furthermore, it introduces a timeliness decay factor and estimated service resource consumption to generate a comprehensive priority index, which is then used to match differentiated response strategies. This system and method can dynamically and accurately quantify the comprehensive processing value of each inquiry, prioritizing the response time of high-value, high-credit customers with limited resources, while simultaneously implementing risk control strategies for low-credit customers. This effectively balances improving conversion rates with managing financial risks, thereby enhancing the overall resource utilization efficiency and profitability of the enterprise.

[0042] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0043] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A business-finance integrated data management system based on big data analytics, characterized in that: include: The inquiry access module is used to receive inquiry requests carrying customer identifiers and inquiry content in real time; The business and financial data fusion module is used to extract business transaction records and financial credit records from the business database and financial database respectively based on the customer identifier, align the business data and financial data according to the timestamp, so that the business records and financial records under the same customer identifier correspond to the same time window, and extract preset feature fields from the aligned data to generate a business and financial fusion feature vector. The customer classification module is used to identify customer types based on the business-finance fusion feature vector, and the customer types include new customers, old customers, healthy business customers, risky business customers, high-credit customers and low-credit customers. The system load awareness module is used to monitor the number of pending inquiries and the average processing rate in the inquiry processing queue in real time, and to calculate the predicted waiting time for the current inquiry. The customer behavior prediction module is used to predict the customer's inquiry abandonment risk function under the current system state by using the business and finance fusion feature vector and the pre-trained survival analysis model, and to calculate the inquiry survival probability based on the predicted waiting time. The value quantification and risk adjustment module is used to predict the benchmark transaction probability and expected gross profit of the inquiry based on the business-finance integration feature vector, the inquiry content and the inquiry survival probability, calculate the expected credit loss combined with financial credit, and obtain the risk-adjusted opportunity value accordingly. The priority calculation module is used to calculate the comprehensive priority index of the inquiry by using the risk-adjusted opportunity value, combined with the timeliness decay factor generated based on the predicted waiting time and the estimated service resource consumption. The strategy scheduling module is used to match differentiated response strategies based on the comprehensive priority index and the customer type, and to allocate inquiries to processing terminals in descending order of the comprehensive priority index.

2. The business-finance integrated data management system based on big data analysis according to claim 1, characterized in that, The business and financial data fusion module further includes: extracting transaction frequency, average transaction amount, return rate, and gross profit contribution from the business side within the most recent preset period, and accounts receivable turnover days, overdue number, aging structure, and external credit rating score from the financial side; filling missing values ​​and standardizing continuous fields; and calculating a weighted average of time series data with greater weight for data closer to the current time, wherein the weight decreases exponentially with historical time intervals, and outputting a business and financial fusion feature vector containing a business vitality index and a financial stability index.

3. The business-finance integrated data management system based on big data analysis according to claim 2, characterized in that, The customer behavior prediction module includes: The model building unit collects customer waiting time, abandonment status, and transaction outcomes from historical inquiry records. Using the business-finance integration feature vector and waiting time as covariates, a Cox proportional hazards model is constructed. The parameters of the Cox proportional hazards model are solved using partial likelihood estimation to obtain the benchmark risk function. and covariate coefficient vector ,in For a baseline instantaneous abandonment risk rate that changes only over time and is independent of covariates, This is a vector consisting of the regression coefficients corresponding to each covariate; The abandonment risk calculation unit is used to, for the current inquiry, substitute its business-finance integration feature vector X into the Cox proportional risk model to obtain the instantaneous abandonment risk rate function. ,in for The inner product of the transpose of X and X; The survival probability calculation unit is used to obtain the predicted waiting time. By analyzing the instantaneous abandonment risk rate function from zero to... Integrate over the interval and take the exponent to calculate the customer's waiting time. The probability of survival if one does not give up within a certain time. The calculation process is expressed as follows: ； Where u is the integration variable, representing time.

4. The business-finance integrated data management system based on big data analysis according to claim 3, characterized in that, The benchmark transaction probability in the value quantification and risk adjustment module It is obtained through a pre-trained gradient boosting tree model, which takes the business-finance fusion feature vector and the product type being inquired about as input features and outputs a predicted transaction probability value; the value quantification and risk adjustment module also calculates the dynamic transaction probability, the dynamic transaction probability... The product of the baseline transaction probability and the inquiry survival probability is expressed as follows: .

5. The business-finance integrated data management system based on big data analysis according to claim 4, characterized in that, The process by which the value quantification and risk adjustment module obtains the risk-adjusted opportunity value includes: Based on the estimated transaction amount A from the inquiry, and using the historical gross profit margin in the business-finance integration feature vector and the current market reference index to predict the current gross profit margin m, where the current market reference index is an external data indicator reflecting industry supply and demand and price fluctuations, the nominal opportunity profit is calculated. ; Using the overdue status transition matrix constructed from the overdue transfer ratios of different aging periods in the financial data, and combined with a pre-trained logistic regression model, the probability of default PD for this transaction is predicted. Combining the risk exposure EAD=A×(1-prepayment ratio) and the preset default loss rate LGD, the expected credit loss ECL=EAD×PD×LGD is calculated, where the prepayment ratio is the proportion of the prepayment required before the transaction to the transaction amount, which is preset according to the customer type. The risk-adjusted opportunity value is obtained by deducting the expected credit loss from the nominal opportunity profit. .

6. The business-finance integrated data management system based on big data analysis according to claim 5, characterized in that, The system load perception module determines a priority adjustment factor θ based on the number of pending inquiries Q in the current queue and the average processing rate μ, and according to the customer type output by the customer classification module. θ for high-credit customers and healthy business customers is set to a first value, while θ for low-credit customers and business-risk customers is set to a second value greater than the first value, determined through a relational formula. Calculate the predicted waiting time.

7. The business-finance integrated data management system based on big data analysis according to claim 6, characterized in that, The priority comprehensive calculation module includes: The time loss calculation unit is used to calculate the predicted waiting time. Compared to the preset average acceptable response time for customers based on industry statistics Comparison, generating the aging decay factor γ, the calculation process is as follows: when Greater than hour, Otherwise γ=1, where This is the decay rate control parameter; The composite index calculation unit is used to calculate the risk-adjusted opportunity value. Multiplying the result by the time-depreciation factor γ and then dividing by the sum of the estimated service resource consumption C and the zero-prevention constant ε, we obtain the comprehensive priority index P. The calculation process is expressed as follows: ； The estimated service resource consumption C is obtained by mapping the preset complexity level of the materials in the inquiry content and the average time taken to process the customer's historical inquiries, and ε is a preset positive number.

8. The business-finance integrated data management system based on big data analysis according to claim 7, characterized in that, The comprehensive index calculation unit also dynamically adjusts the attenuation rate control parameters according to customer type. When a customer is identified as a high-credit customer or a customer with healthy business practices, reduce... The value of is increased when a customer is identified as a high-risk or low-credit customer. The value of .

9. The business-finance integrated data management system based on big data analysis according to claim 8, characterized in that, The strategy scheduling module includes a preemptive scheduling submodule, which is used to pause the processing of the lowest index query when the comprehensive priority index of a new query exceeds the lowest index among all queries being processed and exceeds a preset ratio, retain its processing progress status, update its predicted waiting time to the sum of the waiting time and the waiting time re-estimated by the system load perception module based on the current queue status, recalculate the survival probability and comprehensive priority index, and then insert it back into the queue.

10. A business-finance integrated data management method based on big data analytics, characterized in that: The application of the business-finance integrated data management system based on big data analytics as described in any one of claims 1-9 includes the following steps: Receive inquiry requests in real time, carrying customer identification and inquiry details; Based on the customer identifier, business transaction records and financial credit records are extracted from the business database and financial database respectively. The business data and financial data are aligned by timestamp so that the business records and financial records under the same customer identifier correspond to the same time window. Preset feature fields are extracted from the aligned data to generate a business-finance fusion feature vector. Customer types are identified based on the business-finance integration feature vector, which includes new customers, old customers, healthy business customers, risky business customers, high-credit customers, and low-credit customers. Real-time monitoring of the number of pending inquiries and average processing rate in the inquiry processing queue; calculation of the predicted waiting time for the current inquiry. Using the business-finance fusion feature vector and the pre-trained survival analysis model, the inquiry abandonment risk function of customers in the current system state is predicted, and the inquiry survival probability is calculated based on the predicted waiting time. Based on the business-finance integration feature vector, the inquiry content, and the inquiry survival probability, the benchmark transaction probability and expected gross profit of the inquiry are predicted, the expected credit loss combined with financial credit is calculated, and the risk-adjusted opportunity value is obtained accordingly. Based on the predicted waiting time, a time-depletion factor is generated. The risk-adjusted opportunity value is then adjusted using the time-depletion factor and divided by the estimated service resource consumption to obtain the comprehensive priority index. Differentiated response strategies are matched based on the comprehensive priority index and the customer type, and inquiries are allocated to processing terminals in descending order of comprehensive priority index.

Images