Process monitoring method and device, computer equipment and storage medium

Abstract

The invention relates to a process monitoring method and device, computer equipment and a storage medium. The method comprises the steps that node variables of all nodes in the current process are obtained, the node variables comprise input variables and output variables, statistical analysis is carried out according to the node variables of all the nodes, a statistical result is obtained, and the statistical result comprises distribution information of the node variables and the incidence relation between the different variables; and performing binning processing on the node variable of each node according to the distribution information, and monitoring the node variable of each node according to a binning result and an association relationship among different variables. By adopting the method, the node variables can be monitored based on the binning result and the incidence relation between different variables, and the process monitoring efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of information technology, and in particular, to a process monitoring method, apparatus, computer device, and storage medium. Background Art

[0002] In the context of the widespread application of process engines in today's companies, although many conveniences are brought, a series of technical problems that cannot be ignored have also arisen. In the use of process engines, flowcharts are drawn manually, and there is no expected judgment on the node variables of the flowchart during the execution process, resulting in the process values of process variable parameters being in an unmonitored state, and thus it is very likely to output results that do not meet expectations.

[0003] In dealing with these process-related problems, traditional technologies mainly rely on manual inspection to interpret and verify business flowcharts. After the flowchart is drawn, it is analyzed manually to determine whether the process is reasonable and whether there are errors.

[0004] However, the traditional technology is analyzed manually, which has the problem of low efficiency. Especially in the face of complex business processes, it takes a lot of time for manual understanding and analysis, consuming human and time costs. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide a process monitoring method, apparatus, computer device, and storage medium that can improve monitoring efficiency.

[0006] In a first aspect, this application provides a process monitoring method, including:

[0007] Obtain the node variables of each node in the current process, where the node variables include input variables and output variables;

[0008] Perform statistical analysis based on the node variables of each node to obtain a statistical result, where the statistical result includes the distribution information of the node variables and the correlation relationship between different variables;

[0009] Perform binning processing on the node variables of each node according to the distribution information;

[0010] Monitor the node variables of each node according to the binned results and the correlation relationship between different variables.

[0011] In one embodiment, performing binning processing on the node variables of each node according to the distribution information includes:

[0012] Determine the binning boundaries according to the mean value and quartiles in the distribution information;

[0013] Perform binning processing on the node variables of each node according to the binning boundaries.

[0014] In one embodiment, the process monitoring method further includes:

[0015] Training a prediction model using the historical data of each node;

[0016] Predicting the node variables of each node using the prediction model to obtain prediction results;

[0017] Performing binning processing on the node variables of each node according to the distribution information, including:

[0018] Performing binning processing on the prediction results and the node variables of each node together according to the distribution information.

[0019] In one embodiment, the statistical result further includes outliers. Before training the prediction model using the historical data of each node, the process monitoring method further includes:

[0020] Performing data cleaning on the historical data of each node according to the outliers;

[0021] The method further includes:

[0022] Obtaining the prediction error of the prediction model at the outliers;

[0023] Adjusting the parameters of the prediction model according to the cross-validation algorithm and the prediction error.

[0024] In one embodiment, the process monitoring method further includes:

[0025] When the output variable of the target node is abnormal, obtaining the associated nodes of the target node according to the association relationship between different nodes, and obtaining the associated variables according to the associated nodes;

[0026] Adjusting the associated variables according to the output variable of the target node.

[0027] In one embodiment, when the target node is an approval node, the output variable includes the approval conversion rate, and the process monitoring method further includes:

[0028] When the actual value of the approval conversion rate is lower than the preset threshold, it is determined that the output variable of the target node is abnormal;

[0029] The associated nodes of the approval node include risk conversion nodes, and the associated variable includes a credit score threshold. Adjusting the associated variable according to the output variable of the target node includes:

[0030] Obtaining the actual value of the approval conversion rate and the current credit score threshold;

[0031] Determining a new credit score threshold according to the actual value, the current credit score threshold, and the preset target conversion rate;

[0032] Adjust the current credit score threshold according to the new credit score threshold.

[0033] In one of the embodiments, the process monitoring method further includes:

[0034] Obtain the associated nodes of the risk conversion node, and the associated nodes of the risk conversion node include the quota nodes;

[0035] Obtain the reduction range of the credit score threshold, and obtain the current virtual resource application quota according to the quota node;

[0036] Determine the upper limit of the virtual resource application quota according to the current virtual resource application quota and the reduction range of the credit score threshold;

[0037] Adjust the current virtual resource application quota according to the upper limit of the virtual resource application quota.

[0038] In a second aspect, a process monitoring device is provided, including:

[0039] An acquisition module, configured to acquire the node variables of each node in the current process, where the node variables include input variables and output variables;

[0040] A statistics module, configured to perform statistical analysis according to the node variables of each node to obtain a statistical result, where the statistical result includes the distribution information of the node variables and the correlation relationship between different variables;

[0041] A processing module, configured to perform binning processing on the node variables of each node according to the distribution information;

[0042] A monitoring module, configured to monitor the node variables of each node according to the binned result and the correlation relationship between different variables.

[0043] In a third aspect, the present application provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the process monitoring method provided in any one of the embodiments of the first aspect of the present application are implemented.

[0044] In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the process monitoring method provided in any one of the embodiments of the first aspect of the present application are implemented.

[0045] The above-mentioned process monitoring method, device, computer equipment and storage medium can deeply understand the distribution characteristics of data and the correlation relationships between variables through statistical analysis of node variables, and help discover potential data anomalies and risk points. For example, if an abnormal correlation is found between the input variable and the output variable of certain nodes, it may indicate potential problems or risks in the process. By performing binning processing, the node variables are transformed into categories with a certain range of intervals, making variables with different value ranges comparable and facilitating subsequent analysis and monitoring. At the same time, binning can also perform a certain degree of smoothing processing on the data to reduce the influence of data noise. Further, by monitoring the node variables based on the binning results and the correlation relationships between different variables, abnormal situations can be discovered more efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a flowchart of the process monitoring method in some embodiments;

[0047] Figure 2 is a structural block diagram of the process monitoring device in some embodiments;

[0048] Figure 3 is an internal structure diagram of the computer equipment in some embodiments. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0050] In a first aspect, the present application provides a process monitoring method. As Figure 1 shown, taking the application of this method to a server as an example for illustration, it includes the following steps:

[0051] Step S11, obtain the node variables of each node in the current process, where the node variables include input variables and output variables.

[0052] Among them, the process refers to a business process implemented and managed through a process engine. In the operation of an enterprise or organization, the business process covers a series of interrelated and sequential activities that work together to achieve specific business goals, such as procurement processes, approval processes, production processes, etc.

[0053] A node refers to an operation node in the process, which is used to execute the steps in the process. The current process may include multiple nodes.

[0054] A node variable refers to a variable that can represent certain characteristics, states, or output results of a node in various nodes of a process. For example, in an order processing process, a certain node is responsible for calculating the total order price, and "total order price" can be used as a node variable.

[0055] Specifically, in this application, each node can be pre-configured, and input variables, output variables, and the outlier range and normal value range of each variable can be configured for each node.

[0056] When the process starts and is in the execution process, the server can capture in real time the data generated after each node is executed. This data includes the values of the input variables and output variables of the node.

[0057] The server will deeply monitor the collected execution completion data based on the configuration information pre-configured for each node, that is, input variables, output variables, and the outlier range and normal value range of each variable.

[0058] Step S12: Conduct statistical analysis based on the node variables of each node to obtain statistical results. The statistical results include the distribution information of the node variables and the correlation relationships between different variables.

[0059] Among them, the distribution information refers to the information used to describe the value distribution characteristics of the output variable. Specifically, the distribution information can include values such as mean, standard deviation, and the data distribution form (normal distribution, skewed distribution, etc.).

[0060] The correlation relationship between different variables refers to the relationship of mutual dependence, mutual influence, and mutual restriction existing between variables in a business process, mainly including the following types:

[0061] Data transfer relationship: The value of one variable serves as the input of another variable. For example, in an order processing process, the order amount variable will serve as the input of the tax calculation variable. When the order amount changes, the tax will also change accordingly.

[0062] Causal relationship: A change in one variable will cause a change in another variable. For example, in a production process, a delay in the raw material supply variable will cause a delay in the production progress variable.

[0063] Logical dependence relationship: There is a logical sequence or conditional judgment relationship between variables. For example, in an approval process, only when the approval opinion variable is "approved", the start variable of the next process will be triggered.

[0064] Constraint relationship: The value range or state of one variable will limit the value or state of another variable. For example, in an inventory management process, the inventory quantity variable will limit the order deliverable quantity variable, and the order deliverable quantity cannot exceed the inventory quantity.

[0065] Parallel relationship: Multiple variables affect or act together during the same time or stage. For example, in the project management process, the task progress variable and the resource allocation variable may affect the overall progress of the project simultaneously during the project execution stage.

[0066] Specifically, this application can use statistical methods to study the data distribution. For example, by analyzing the order amount data to see whether it is concentrated in a certain price range (such as concentrated between 100 - 200 yuan) or evenly distributed across different price intervals. By judging whether the data conforms to the expected pattern, such as whether it conforms to the normal distribution, to evaluate whether the process execution is normal. If the expected order amount conforms to the normal distribution, but the actual analysis finds a large deviation, it may indicate that there are problems in the process, such as incorrect price calculation or abnormal orders.

[0067] Step S13: Perform binning on the node variables of each node according to the distribution information.

[0068] Among them, binning is a data processing technique mainly used to divide continuous variables or discrete variables into a finite number of "bins" or intervals for operations such as data analysis and modeling. The purpose of binning is to discretize the data, that is, to convert continuous numerical variables into discrete categorical variables, making the data easier to understand and process. Especially in some algorithms, discrete variables may be more convenient for calculation and analysis. In addition, binning can also reduce data noise. By merging adjacent data values into the same bin, the minor fluctuations and noise in the data can be reduced, making the overall characteristics of the data more obvious.

[0069] Specifically, this application can perform binning operations on the data related to the collected node variables. This operation aims to divide the continuous data into different intervals according to certain rules to better observe the distribution characteristics and change trends of the data. For example, according to factors such as the size range of the data and the frequency of the data, the data is reasonably divided into different bins.

[0070] Furthermore, using the data configurations after analysis and binning, continuously monitor the node variables during the process execution in all aspects. Once abnormal data is found, such as deviating from the preset normal value range or having a large difference from the historical data pattern, the system will immediately issue an alarm so that relevant personnel can take measures in a timely manner to make adjustments to ensure that the output result of the entire process meets the expectations.

[0071] Specifically, the binning process of this application can specifically include:

[0072] Equal-width binning: Divide the value range of a variable into several intervals with equal widths. For example, divide the age variable into intervals of every 10 years, with 0 - 10 years in one bin, 11 - 20 years in another bin, and so on.

[0073] Equal-frequency binning: Also known as equal-depth binning, it makes the number of data points in each bin approximately equal. For example, divide 100 data points into 5 bins, with each bin containing approximately 20 data points.

[0074] Clustering-based binning: Use clustering algorithms such as K-Means to divide data points into different bins according to their similarity, and similar data points will be assigned to the same bin.

[0075] Step S14: Monitor the node variables of each node according to the binned results and the correlation relationships between different variables.

[0076] Specifically, this application can determine monitoring indicators. Based on the data of the output variables after binning, determine indicators that can reflect the characteristics of the variables, such as the frequency, frequency, mean, median, etc. of each bin. Combine the correlation relationships between variables to determine comprehensive indicators, such as the correlation coefficient and conditional probability between related variables, to measure the tightness of the relationship between variables.

[0077] Furthermore, set reasonable thresholds for the monitoring indicators according to business experience, historical data, or statistical methods. Set the upper and lower limits of abnormal frequencies for the frequency after binning; set the normal range threshold for the correlation coefficient of related variables. When the indicator exceeds the threshold, it is regarded as an abnormal situation.

[0078] Furthermore, calculate the monitoring indicators after binning and analyze whether the correlation relationships between variables have changed. Compare the real-time calculated monitoring indicators with the set thresholds. If the indicator exceeds the threshold range, trigger an abnormal alarm. At the same time, analyze whether the abnormality is caused by an abnormal single variable or an abnormal correlation relationship between variables.

[0079] Furthermore, adjust the business process or model parameters in a timely manner according to the monitoring results. If it is found that the correlation relationship between variables has changed, re-evaluate the model assumptions and parameter settings; if an abnormal situation occurs, deeply analyze the reasons and take corresponding measures to solve them.

[0080] In one embodiment, binning processing is performed on the node variables of each node according to the distribution information, including: determining the binning boundaries according to the mean and quartiles in the distribution information, and performing binning processing on the node variables of each node according to the binning boundaries.

[0081] Among them, the mean, that is, the average, is the sum of a set of data divided by the number of data, which reflects the central tendency of the data.

[0082] The quartiles are also called the four - fractiles, which are three dividing points that divide the data into four equal parts after sorting the data from smallest to largest. The first quartile (Q1) is the value at the 25% position of the lower part of the data, the second quartile (Q2), which is the median, is the value at the middle position of the data, and the third quartile (Q3) is the value at the 25% position of the upper part of the data.

[0083] Specifically, this application can use the mean as a reference point and then combine the quartiles to determine the bin boundaries. For example, a common method is to divide the data range from the minimum value to the maximum value into multiple intervals, using Q1, the mean, Q3, etc. as the key bin boundary points. For instance, the part less than Q1 can be taken as one bin, the part from Q1 to the mean as one bin, the part from the mean to Q3 as one bin, and the part greater than Q3 as one bin. Of course, the specific division method can be adjusted according to the data characteristics and analysis purposes.

[0084] Further, after determining the bin boundaries, the values of the output variable are classified into the corresponding bins according to their numerical magnitudes. For example, if there is an output variable with a value of 5 and the bin boundaries are [0, 3), [3, 7), [7, 10], then 5 will be classified into the bin [3, 7).

[0085] The beneficial effect of this embodiment is that a large number of data points are merged into a limited number of bins, which simplifies the data representation and facilitates understanding the distribution characteristics of the data. Binning can smooth the data, reduce the influence of individual extreme values or noise on the analysis, and make the overall trend of the data more obvious.

[0086] In one of the embodiments, the process monitoring method may further include: training a prediction model using the historical data of each node, predicting the node variables of each node using the prediction model to obtain prediction results, and performing binning processing on the node variables of each node according to the distribution information, including: performing binning processing on the prediction results and the node variables of each node together according to the distribution information.

[0087] Specifically, this application can, according to the historical data and the selected prediction model (such as linear regression, decision tree, etc.), learn the patterns and rules in the data by training the model, and then predict the result data of the current node to obtain the possible future development trends or results. The prediction result, that is, the value or trend predicted by the model, such as predicting the production output, task completion time, etc. in the next time period.

[0088] Further, the prediction results can be used as the basis for binning together with the original execution result data, which can reflect the data characteristics from a more comprehensive perspective. For example, binning after combining the prediction results and the actual execution results helps to observe the distribution of the predicted values and the actual values in different intervals.

[0089] The beneficial effects of this embodiment are as follows: By observing the distribution of predicted values and actual values in different intervals, the prediction accuracy of the model in different data segments can be intuitively understood, so as to comprehensively evaluate the performance of the model in various situations, facilitate the discovery of the advantages and disadvantages of the model, and then improve the model targeted. Merging bins can reveal the potential relationship between the prediction results and the actual results, help analyze the changing trends and laws of data in different feature intervals, provide more dimensional information for further data analysis, and contribute to the discovery of some hidden associations and patterns in the business.

[0090] In one of the embodiments, the statistical results further include outliers. Before training the prediction model using the historical data of each node, the process monitoring method may further include: cleaning the historical data of each node according to the outliers, and the process monitoring method may further include: obtaining the prediction error of the prediction model at the outliers, and adjusting the parameters of the prediction model according to the cross-validation algorithm and the prediction error.

[0091] Among them, an outlier refers to an observed value or data point in a data set that is significantly different from other data points and deviates from the overall data distribution pattern.

[0092] The cross-validation algorithm is a technique for evaluating and optimizing machine learning models. It divides the data set into different subsets and conducts multiple trainings and validations on these subsets to more accurately evaluate the performance and generalization ability of the model. A common implementation method is K-fold cross-validation.

[0093] Specifically, this application can use outlier detection methods such as based on statistics, distance or density to identify outliers in the historical data. Input the outlier data into the trained prediction model to obtain the prediction results, and then calculate the difference from the actual outlier values. Usually, metrics such as mean squared error and mean absolute error are used to quantify the prediction error.

[0094] Furthermore, the data set is divided into multiple subsets, usually using K-fold cross-validation. Each time, one of the subsets is used as the validation set, and the remaining subsets are used as the training set. Train and evaluate the performance of the model on different subsets to obtain multiple prediction error results.

[0095] According to the prediction error obtained by cross-validation, use an optimization algorithm such as the gradient descent method to adjust the parameters of the prediction model. For example, if it is found that the prediction error of the model at the outliers is large, adjust the parameters to make the prediction of the model closer to the true value at these points, while trying to keep the performance at other normal data points from decreasing.

[0096] The beneficial effects of this embodiment are as follows: By adjusting the model parameters for outliers, the model can better handle abnormal situations, reduce the impact of outliers on the prediction results, and thus improve the overall prediction accuracy and reliability. After processing the outliers and adjusting the parameters, the model has a stronger tolerance for noise and outliers in the data. When facing data with various different distributions, it can more stably output accurate prediction results, improving the robustness of the model.

[0097] In one embodiment, the process monitoring method may further include: When the output variable of the target node is abnormal, obtaining the associated nodes of the target node according to the association relationship between different variables, and obtaining the associated variables according to the associated nodes, and adjusting the associated variables according to the output variable of the target node.

[0098] Among them, an associated node refers to other nodes in the current process that have a direct or indirect relationship with a specific node (such as the target node) in terms of data transfer, business logic, or process execution. The status or output of these nodes will affect the target node or be affected by the target node.

[0099] An associated variable refers to an input or output variable related to an associated node. They are the data transferred between nodes, and through these variables, information interaction and dependence are achieved between nodes. For example, the output variable of one node may be the input variable of another node, then this variable is an associated variable for these two nodes.

[0100] Specifically, this application can determine whether the output variable is abnormal by setting a threshold. Specifically, the mean and standard deviation of the output variable under normal conditions can be calculated based on historical data. When the current output variable exceeds the range of the mean plus or minus several times the standard deviation, it is determined to be abnormal.

[0101] When it is detected that the output variable of the target node is abnormal, according to the association relationship between different variables, find the nodes that have an association relationship with the target node. If the association relationship is represented by a graph structure, traverse the edges connected to the target node in the graph to find the corresponding associated nodes. If the association relationship is recorded through a data dictionary or configuration file, directly find the node information related to the target node from it.

[0102] For each associated node, obtain the corresponding associated variable according to its definition in the association relationship. If the input variable of the associated node has a data transfer relationship with the output variable of the target node, then this input variable is an associated variable; similarly, if the output variable of the associated node will affect the target node, this output variable is also an associated variable. The values of these associated variables can be extracted from the input and output data records of the node.

[0103] Further, analyze the anomalies of the output variables of the target node to determine the adjustment strategy. For example, if the output variable of the target node is too high, it may be necessary to reduce the value of the associated variable related to it; if the output variable of the target node is too low, it may be necessary to increase the value of the relevant associated variable. Adjust the associated variable according to the determined adjustment strategy. Methods such as linear adjustment and proportional adjustment can be used, or more complex adjustment methods based on rules or machine learning can also be used. After adjustment, feedback the new value of the associated variable to the corresponding associated node, continue to monitor the running status of the process, and observe whether the output variable of the target node returns to normal.

[0104] The beneficial effects of this embodiment are as follows: when an anomaly occurs in the target node, it is possible to quickly locate the associated nodes and variables related to it, which helps to quickly identify the root cause of the problem and improve the efficiency of fault diagnosis and handling. By adjusting the associated variable, more refined control and optimization of the entire process can be achieved. Since the associated variable reflects the mutual influence relationship between nodes, reasonable adjustment of them can make the process run more stably and efficiently, and avoid the entire process from malfunctioning or degrading in performance due to the anomaly of a certain node.

[0105] In one embodiment, when the target node is an approval node, the output variable includes the approval conversion rate. The process monitoring method may further include: when the actual value of the approval conversion rate is lower than the preset threshold, determine that the output variable of the target node is abnormal. The associated node of the approval node includes a risk conversion node, and the associated variable includes a credit score threshold. Adjusting the associated variable according to the output variable of the target node includes: obtaining the actual value of the approval conversion rate and the current credit score threshold, determining a new credit score threshold according to the actual value, the current credit score threshold, and the preset target conversion rate, and adjusting the current credit score threshold according to the new credit score threshold.

[0106] Among them, the approval node is mainly used to review various information and operations submitted in the process to ensure that they meet the requirements of relevant rules and regulations, business processes, and laws and regulations. For example, in the loan application process, the approval node will review whether the materials submitted by the applicant are complete and whether the application process is correct.

[0107] The risk conversion node is mainly used to comprehensively identify various risks that may exist in the process by using professional risk assessment models and methods, such as credit risk, market risk, operation risk, etc. For example, in the investment project process, the risk conversion node will analyze risk factors such as the market prospect, competition status, and policy and regulation changes of the project.

[0108] A credit score threshold is a specific score boundary used to divide different credit levels or decision results in a credit assessment system. Financial institutions or other relevant entities will set different credit score thresholds based on factors such as their own business needs, risk tolerance, and market conditions. For example, when the credit score is higher than a certain threshold, a loan application may be approved, a higher credit limit may be granted, or more favorable financial services may be provided; while when the credit score is lower than the threshold, the loan application may be rejected, the credit limit may be restricted, or additional collateral may be required, etc.

[0109] Specifically, to determine a new credit score threshold based on the actual value, the current credit score threshold, and a pre-set target conversion rate, it can be calculated using the following formula:

[0110] Approval conversion rate = Number of approved applications / Total number of applications × 100%;

[0111] New credit score threshold = Current credit score threshold - (Target conversion rate - Actual value) × Coefficient;

[0112] Among them, the coefficient can be pre-configured according to the business risk preference. For example, it can be taken as 100.

[0113] Exemplarily, assume that the current credit score threshold is 600 points, the target conversion rate is 60%, and when the actual conversion rate is 50%, the new credit score threshold = 600 - (60% - 50%) × 100 = 500 points. Then the server will lower the current credit score threshold from 600 points to 500 points.

[0114] The beneficial effect of this embodiment is that it monitors the output variable of the approval node, the approval conversion rate. When an abnormal approval conversion rate is monitored, it obtains the associated node of the approval node, that is, the risk conversion node, and obtains the associated variable and the credit score threshold. By adjusting the credit score threshold, it realizes improving the approval conversion rate and realizes the associated adjustment between different nodes.

[0115] In one of the embodiments, the process monitoring method may further include: obtaining the associated node of the risk conversion node. The associated node of the risk conversion node includes the quota node, obtaining the reduction range of the credit score threshold, and obtaining the current virtual resource application quota according to the quota node.

[0116] Determine the upper limit of the virtual resource application quota according to the current virtual resource application quota and the reduction range of the credit score threshold, and adjust the current virtual resource application quota according to the upper limit of the virtual resource application quota.

[0117] Among them, the quota node is mainly used to determine the quota standards for various types of services and perform quota adjustments based on relevant business rules and factors such as the customer's credit status and financial status.

[0118] The virtual resource application quota can be the loan quota, and the upper limit of the virtual resource application quota can be the upper limit of the loan quota.

[0119] Determine the upper limit of the virtual resource application quota according to the current virtual resource application quota and the decrease rate of the credit score threshold, which can be calculated by the following formula:

[0120] Upper limit of virtual resource application quota = Original virtual resource application quota × (1 - Decrease rate ratio of credit score threshold)

[0121] For example, if the credit score threshold drops from 600 to 500, the decrease in the credit score threshold is 16.7%. Substitute the current virtual resource application quota to calculate the upper limit of the virtual resource application quota.

[0122] Furthermore, compare the virtual resource application quota with the upper limit of the virtual resource quota. If the virtual resource application quota is lower than or equal to the upper limit, usually no adjustment is required and it can directly enter the subsequent approval process.

[0123] If the virtual resource application quota is higher than the upper limit, the virtual resource application quota needs to be adjusted to the upper limit.

[0124] In one embodiment, the process monitoring method may further include: obtaining the original coefficient of the interest rate according to the quota node, determining the interest rate floating coefficient according to the original coefficient of the interest rate and the decrease rate of the credit score threshold, and adjusting the interest rate according to the interest rate floating coefficient.

[0125] Among them, the interest rate floating coefficient refers to a numerical coefficient used to adjust the benchmark interest rate to determine the actual executed interest rate in interest rate pricing.

[0126] Specifically, determining the interest rate floating coefficient according to the original coefficient of the interest rate and the decrease rate of the credit score threshold can be calculated according to the following formula:

[0127] Interest rate floating coefficient = Original coefficient × (1 + Decrease rate ratio of credit score threshold)

[0128] It can be seen that this application can jointly adjust the credit score threshold, the upper limit of the virtual resource application quota, and the interest rate floating coefficient when the approval conversion rate in the approval node is abnormal, realizing multi-parameter linkage control.

[0129] In a second aspect, this application provides a process monitoring device, as Figure 2 shown. The process monitoring device includes: an acquisition module 21, a statistics module 22, a processing module 23, and a monitoring module 24, where:

[0130] The acquisition module 21 is used to acquire the node variables of each node in the current process, and the node variables include input variables and output variables;

[0131] A statistical module 22 for performing statistical analysis based on the node variables of each node to obtain a statistical result, where the statistical result includes the distribution information of the node variables and the correlation relationship between different variables;

[0132] A processing module 23 for performing binning processing on the node variables of each node according to the distribution information;

[0133] A monitoring module 24 for monitoring the node variables of each node according to the binned result and the correlation relationship between different variables.

[0134] In some embodiments, the above-mentioned processing module 23 may determine the binning boundaries according to the mean value and quartiles in the distribution information, and perform binning processing on the node variables of each node according to the binning boundaries.

[0135] In some embodiments, the above-mentioned processing module 23 may also train a prediction model using the historical data of each node, use the prediction model to predict the node variables of each node to obtain a prediction result, and perform binning processing on the prediction result and the node variables of each node together according to the distribution information.

[0136] In some embodiments, the above-mentioned statistical result further includes outliers. The above-mentioned processing module 23 may also perform data cleaning on the historical data of each node according to the outliers, obtain the prediction error of the prediction model at the outliers, and adjust the parameters of the prediction model according to the cross-validation algorithm and the prediction error.

[0137] In some embodiments, when the output variable of the target node is abnormal, the above-mentioned monitoring module 24 may obtain the associated nodes of the target node according to the correlation relationship between different nodes, obtain the associated variables according to the associated nodes, and adjust the associated variables according to the output variable of the target node.

[0138] In some embodiments, when the above-mentioned target node is an approval node, the output variable includes the approval conversion rate. When the actual value of the approval conversion rate is lower than the preset threshold, the above-mentioned monitoring module 24 may determine that the output variable of the target node is abnormal. The associated nodes of the approval node include risk conversion nodes, and the associated variable includes the credit score threshold. The above-mentioned monitoring module 24 may also obtain the actual value of the approval conversion rate and the current credit score threshold, determine a new credit score threshold according to the actual value, the current credit score threshold, and the preset target conversion rate, and adjust the current credit score threshold according to the new credit score threshold.

[0139] In some of these embodiments, the above-mentioned monitoring module 24 may further obtain the associated nodes of the risk transformation node. The associated nodes of the risk transformation node include quota nodes. It obtains the reduction in the credit score threshold, and obtains the current virtual resource application quota according to the quota node. It determines the upper limit of the virtual resource application quota according to the current virtual resource application quota and the reduction in the credit score threshold, and adjusts the current virtual resource application quota according to the upper limit of the virtual resource application quota.

[0140] In a third aspect, the present application provides a computer device. The computer device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the process monitoring method provided in any of the embodiments of the first aspect of the present application.

[0141] In one embodiment, the computer device may be a server, and its internal structure diagram may be as Figure 3 shown. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program, when executed by the processor, implements the process monitoring method.

[0142] In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the process monitoring method provided in any of the embodiments of the first aspect of the present application.

[0143] The computer-readable storage medium may be the Figure 3 computer-readable storage medium in the computer device shown.

[0144] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The above computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above various methods. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in this application can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0145] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0146] The above embodiments only represent several implementation manners of this application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of the patent of this application should be subject to the appended claims.

Claims

1. A process monitoring method, characterized in that: The method comprises: Obtain node variables of each node in the current process, wherein the node variables include input variables and output variables; Performing statistical analysis on the node variables of each of the nodes to obtain statistical results, wherein the statistical results include distribution information of the node variables and correlation relationships between different variables; Perform binning processing on the node variables of each of the nodes according to the distribution information; The node variables of each node are monitored according to the binning results and the correlation between the different variables.

2. The method according to claim 1, characterized in that The binning of the node variables of each of the nodes according to the distribution information includes: Determine the boundaries of the bins according to the mean and the four-dimensional digits in the distribution information; The node variables of each of the nodes are binned according to the boundaries of the bins.

3. The method according to claim 1, characterized in that The method further comprises: Using the historical data of each of the nodes to train a prediction model; Using the prediction model to predict the node variables of each node to obtain a prediction result; The binning of the node variables of each of the nodes according to the distribution information includes: The prediction result and the node variables of each of the nodes are binned according to the distribution information.

4. The method according to claim 3, characterized in that The statistical results also include outliers. Before using the historical data of each of the nodes to train the prediction model, the method further includes: Performing data cleaning on the historical data of each of the nodes according to the outliers; The method further comprises: Obtaining the prediction error of the prediction model at the outlier point; The parameters of the prediction model are adjusted according to the cross-validation algorithm and the prediction error.

5. The method according to claim 1, characterized in that The method further comprises: When an output variable of a target node is abnormal, an associated node of the target node is obtained according to the association relationship between the different nodes, and an associated variable is obtained according to the associated node; The associated variable is adjusted according to the output variable of the target node.

6. The method according to claim 5, characterized in that When the target node is an approval node, the output variable includes an approval conversion rate, and the method further includes: When the actual value of the approval conversion rate is lower than a preset threshold, it is determined that an output variable of the target node is abnormal; The associated node of the approval node includes a risk conversion node, the associated variable includes a credit score threshold, and adjusting the associated variable according to the output variable of the target node includes: Get the actual value of the approval conversion rate and the current credit score threshold; Determine a new credit score threshold according to the actual value, the current credit score threshold and a preset target conversion rate; The current credit score threshold is adjusted according to the new credit score threshold.

7. The method according to claim 3, characterized in that The method further comprises: Acquire the associated nodes of the risk conversion node, where the associated nodes of the risk conversion node include a quota node; Obtaining a decrease in the credit score threshold, and obtaining a current virtual resource application quota according to the quota node; Determining an upper limit of the virtual resource application amount according to the current virtual resource application amount and the decrease in the credit score threshold; The current virtual resource application amount is adjusted according to the upper limit of the virtual resource application amount.

8. A process monitoring device, characterized in that: The device comprises: An acquisition module is used to acquire node variables of each node in the current process, wherein the node variables include input variables and output variables; A statistical module, used for performing statistical analysis on the node variables of each of the nodes to obtain statistical results, wherein the statistical results include distribution information of the node variables and correlation relationships between different variables; A processing module, used for performing binning processing on the node variables of each of the nodes according to the distribution information; The monitoring module is used to monitor the node variables of each of the nodes according to the results after binning and the association relationship between the different variables.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images