Product determination method, device, equipment and computer storage medium
Through neural network training, the exit model is calculated and the characteristics are screened, which solves the problem of long-term and low accuracy of the product exit process in the prior art, and achieves fast and accurate product exit decisions.
Patent Information
- Application Number
- CN202111546709.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-12-16
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2041-12-16
AI Technical Summary
The prior art takes a long time and has low accuracy when determining products that need to be withdrawn, so it is impossible to effectively and efficiently identify products with poor performance.
By obtaining the target product data, training the exit model based on the neural network, calculating the score of the target product, and determining whether it is an exit product based on the preset threshold, using the variable weight formula to determine the feature weight coefficient, perform feature importance screening and model training, and improve accuracy Sex and efficiency.
It realizes the rapid and accurate determination of the products that need to be withdrawn, improves the accuracy of the products that are withdrawn, and reduces the time-consuming process of determination.
Smart Images

Figure CN114219307B_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of big data, and in particular, relates to a product determination method, device, equipment and computer storage medium. Background Art
[0002] With the rapid development of the economy, in order to obtain the effectiveness of product launches, it is necessary to withdraw products with poor results and retain products with better results. Based on the above purpose, the industry has an increasingly strong business demand for product withdrawal mechanisms. At present, the method of determining the products with poor results that need to be withdrawn from existing products mainly relies on relevant technical personnel. The relevant technical personnel analyze the data of the products obtained, determine the products with poor results, and withdraw them, which is time-consuming and has low accuracy. Summary of the invention
[0003] The embodiments of the present application provide a product determination method, apparatus, device, computer storage medium, and computer program product, which can more quickly and accurately determine products that need to be withdrawn.
[0004] In a first aspect, an embodiment of the present application provides a product determination method, the method comprising:
[0005] Acquire target product data, the target product data including a target product feature type and product feature data corresponding to the target product feature type;
[0006] The target score of the target product is calculated based on the exit model and the target product data. The exit model is trained by the product data. The product data includes the product feature type, the data corresponding to the product feature data type and the product exit data. The exit model calculates the target score based on the product feature data corresponding to the target product feature type.
[0007] When the target score of the target product is greater than a first preset threshold, the target product is determined to be an exit product.
[0008] According to one aspect of the present application, before calculating the target score of the target product based on the exit model and the target product data, the method further includes:
[0009] Get product data;
[0010] Train the neural network based on product data to obtain a first model;
[0011] Input product data into the first model to obtain importance scores of product feature types;
[0012] Select product feature types with importance scores greater than a preset threshold as target feature types;
[0013] Filter product data to obtain product feature data corresponding to the target feature type;
[0014] The first model is trained based on the product feature data and product exit data corresponding to the target feature to obtain an exit model.
[0015] According to one aspect of the present application, training a first model based on product feature data and product exit data corresponding to a target feature to obtain an exit model includes:
[0016] The target weight coefficient of the target feature is determined according to the variable weight formula. The variable weight formula is:
[0017]
[0018] in, is the target weight coefficient of the i-th target feature, w i is the weight coefficient of the i-th target feature in the exit model, x i is the evaluation value of the i-th target feature, n is the number of target features, i = 1, 2, ..., n;
[0019] Based on the target weight coefficient of the target feature, the target product feature data is calculated to obtain the target score of the target product.
[0020] According to one aspect of the present application, training a neural network based on product data to obtain a first model includes:
[0021] The product data is evenly divided into k groups, with 1 group as the test sample and k-1 groups as the training samples, where k is greater than or equal to 2;
[0022] Train the neural network based on the training samples to obtain an initial model;
[0023] The initial model is verified based on the verification sample to obtain the first model.
[0024] According to one aspect of the present application, a neural network is trained based on a training sample to obtain an initial model; and the initial model is verified based on a verification sample to obtain a first model, including:
[0025] Normalize the training sample data and the validation sample data;
[0026] Discretize the training sample data and the validation sample data;
[0027] The neural network is trained based on the processed training sample data to obtain an initial model;
[0028] The initial model is verified based on the processed verification sample data to obtain a first model.
[0029] According to one aspect of the present application, the product feature type includes at least one of performance, customer base, market, management attributes, risk data and product attribute data.
[0030] According to one aspect of the present application, before training the neural network based on the processed training sample data to obtain the initial model, the method further includes:
[0031] The null value processing is performed on the sample data and the validation data based on the mean value filling method.
[0032] In a second aspect, an embodiment of the present application provides a product determination device, the device comprising:
[0033] An acquisition module, used to acquire target product data, the target product data including a target product feature type and product feature data corresponding to the target product feature type;
[0034] a calculation module, configured to calculate a target score for a target product based on an exit model and target product data, wherein the exit model is trained by product data, the product data includes product feature types, data corresponding to product feature data types, and product exit data, and the exit model calculates the target score based on the product feature data corresponding to the target product feature type;
[0035] The determination module is used to determine that the target product is an exit product when the target score of the target product is greater than a first preset threshold.
[0036] According to one aspect of the present application, the device further includes:
[0037] The acquisition module is also used to obtain product data;
[0038] A training module, used for training a neural network based on product data to obtain a first model;
[0039] An input module, used for inputting product data into the first model to obtain importance scores of product feature types;
[0040] A selection module, used for selecting a product feature type with an importance score greater than a preset threshold as a target feature type;
[0041] A screening module is used to screen product data and obtain product feature data corresponding to a target feature type;
[0042] The training module is also used to train the first model based on the product feature data and product exit data corresponding to the target feature to obtain the exit model.
[0043] According to one aspect of the present application, the training module is further used to train the first model based on the product feature data and product exit data corresponding to the target feature to obtain an exit model, including:
[0044] The determination module is also used to determine the target weight coefficient of the target feature according to the variable weight formula. The variable weight formula is
[0045]
[0046] in, is the target weight coefficient of the i-th target feature, w i is the weight coefficient of the i-th target feature in the exit model, x i is the evaluation value of the i-th target feature, n is the number of target features, i = 1, 2, ..., n;
[0047] The training module is also used to calculate the target product feature data based on the target weight coefficient of the target feature to obtain the target score of the target product.
[0048] According to one aspect of the present application, a training module, used to train a neural network based on product data to obtain a first model, includes:
[0049] A partitioning module is used to evenly divide the product data into k groups, where one group is used as a test sample and k-1 groups are used as training samples, where k is greater than or equal to 2;
[0050] A training module is used to train the neural network based on the training samples to obtain an initial model;
[0051] The verification module is used to verify the initial model based on the verification sample to obtain the first model.
[0052] According to one aspect of the present application, a training module is used to train a neural network based on a training sample to obtain an initial model; and a verification module is used to verify the initial model based on a verification sample to obtain a first model, including:
[0053] A processing module, used for normalizing the training sample data and the verification sample data;
[0054] The processing module is also used to discretize the training sample data and the verification sample data;
[0055] A training module is used to train the neural network based on the processed training sample data to obtain an initial model;
[0056] The verification module is used to verify the initial model based on the processed verification sample data to obtain the first model.
[0057] According to one aspect of the present application, the product feature type includes at least one of performance, customer base, market, management attributes, risk data and product attribute data.
[0058] According to one aspect of the present application, the device further includes:
[0059] The processing module is also used to perform null value processing on sample data and verification data based on the mean value filling method.
[0060] In a third aspect, an embodiment of the present application provides a product determination device, the device comprising:
[0061] a processor, and a memory storing computer program instructions;
[0062] The processor reads and executes computer program instructions to implement the product determination method of the first aspect.
[0063] In a fourth aspect, an embodiment of the present application provides a computer storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the product determination method of the first aspect is implemented.
[0064] In a fifth aspect, an embodiment of the present application provides a computer program product, characterized in that when instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the product determination method of the first aspect.
[0065] The product determination method, apparatus, device and computer storage medium of the embodiments of the present application obtain target product data, calculate the importance of product features based on the target product data and a neural network, and further determine important features, train the neural network based on the target product data corresponding to the important features to obtain an exit model, calculate the product score using the trained exit model, and use the target products with scores greater than a first preset threshold as exit products, so as to accurately determine the products that need to be exited in a short time. BRIEF DESCRIPTION OF THE DRAWINGS
[0066] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.
[0067] Figure 1 It is a flowchart of a product determination method provided in an embodiment of the present application;
[0068] Figure 2 is a structural schematic diagram of a product determination device provided in an embodiment of the present application;
[0069] Figure 3 It is a structural diagram of a product determination device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0070] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.
[0071] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.
[0072] Currently, the exit of financial products requires relevant technical personnel to analyze the obtained product data and determine the products that need to be exited, which is time-consuming and has low accuracy.
[0073] In order to solve the problems of the prior art, the embodiments of the present application provide a product determination method, device, equipment and computer storage medium. The product determination method provided by the embodiments of the present application is first introduced below.
[0074] Figure 1 FIG. 1 is a flow chart of a product determination method provided by an embodiment of the present application. Figure 1 As shown, the method may include the following steps:
[0075] S110. Acquire target product data, where the target product data includes a target product feature type and product feature data corresponding to the target product feature type.
[0076] The target products acquired are financial products, such as credit card products.
[0077] In some embodiments, the feature type of the target product is obtained, wherein the target product feature type may include product attribute information, performance, customer base, market, management attributes, risk data, etc.; the product feature data corresponding to the target product feature type is obtained as shown in Table 1, for example, including product identification (ASPD_ID), product sales start time (Strt_Sale_Dt), product sales end time (End_Sale_Dt), etc. When the number of target products obtained is 473, the number of corresponding product feature data obtained is also 473.
[0078] Table 1
[0079]
[0080]
[0081] In some embodiments, a feature system library is established based on the mapping relationship between the target product feature type and the product feature data corresponding to the target product feature type, for example: ASPD_ID corresponds to the attribute information of the product.
[0082] S120. Calculate a target score for the target product based on an exit model and target product data. The exit model is trained by product data. The product data includes product feature types, data corresponding to product feature data types, and product exit data. The exit model calculates the target score based on the product feature data corresponding to the target product feature type.
[0083] Input the target product data into the trained exit model to obtain the score of the output target product, wherein the exit model is trained using product data, and the product data includes product feature type, data corresponding to the product feature data type, and product exit data, and the product exit data is used to indicate whether the product is an exit product. The exit model calculates based on the product feature data corresponding to the input target product feature type and outputs the score of the target product. Using the trained exit model, the acquired target product data is input into the model to obtain the output score of the target product, which is used to determine whether the target product needs to be exited.
[0084] S130: When the target score of the target product is greater than a first preset threshold, determine that the target product is an exit product.
[0085] The target scores of the target products are compared, and when the target scores of the target products are greater than a first preset threshold, the product is determined to be an exit product, that is, exit processing is required. The first preset threshold can be set according to demand and is not limited to this.
[0086] In one example, the first preset threshold may be 0.90, and the target score of the product is 0.86. At this time, the product is determined to be an exit product.
[0087] The product determination method provided in the embodiment of the present application can calculate the target product score based on the target product data and the exit model, compare the obtained target product score with the first preset threshold, and determine the exit product, thereby improving the accuracy of the exit product and reducing the time spent in determining the exit product.
[0088] In some embodiments, before calculating the target score of the target product based on the exit model and the target product data, the method also includes: obtaining product data; training a neural network based on the product data to obtain a first model; inputting the product data into the first model to obtain an importance score of the product feature type; selecting a product feature type whose importance score is greater than a preset threshold as the target feature type; screening the product data to obtain product feature data corresponding to the target feature type; training the first model based on the product feature data corresponding to the target feature and the product exit data to obtain an exit model. Obtain product data, the product data including product feature types, data corresponding to product feature data types, and product exit data. Use the product data to train a neural network based on an XGBoost classifier to obtain a first model. Output the product data to the first model as a product feature classification category tree, and calculate the importance score of the product feature type based on the obtained product feature classification category tree. Use the product feature type whose importance score is greater than the preset threshold as the target feature type, screen the product data, delete the product feature types and corresponding product feature data other than the target feature type, and use the screened product data to train the first model to obtain an exit model. The importance score The calculation formula is as follows:
[0089]
[0090]
[0091] x i Characterized by w q(x) is the score of the leaf node q, f(x) is one of the regression trees, R is all the trees, T is the number of leaf nodes, and w is the weight of the leaf node.
[0092] In one example, the feature importance score results are shown in Table 2, idcst_num scores 0.62; actv_cst_num scores 0.24, and com_rel_cst_num_new_rto_qtr scores 0.05.
[0093] Table 2
[0094]
[0095]
[0096] The product determination method provided in the embodiment of the present application can use the product data of the determined exit data for model training, screen the data according to the importance of product features calculated by the model, and further train the model using features with higher importance, so that the product scores output by the model are more accurate, and the accuracy of determining the exited products is higher.
[0097] In some embodiments, training the first model based on the product feature data and product exit data corresponding to the target feature to obtain the exit model includes: determining the weight coefficient of the target feature according to a variable weight formula; training the first model based on the product feature data, product exit data and the weight coefficient of the target feature corresponding to the target feature to obtain the exit model. For the determined target feature, using the variable weight formula:
[0098]
[0099] in, is the target weight coefficient of the i-th target feature; x i is the evaluation value of the i-th target feature; n is the number of target features; i = 1, 2, ..., n; w i is the weight coefficient of the i-th target feature in the exit model. Based on the target weight coefficient of the target feature, the target product feature data is calculated to obtain the target score of the target product.
[0100] In some embodiments, training a neural network based on product data to obtain a first model includes: evenly dividing the product data into k groups, wherein group 1 is used as a test sample and group k-1 is used as a training sample, wherein k is greater than or equal to 2; training the neural network based on the training sample to obtain an initial model; and validating the initial model based on the validation sample to obtain the first model. Based on k-fold cross validation, the product data is divided into k groups, wherein group 1 is used as a test sample and group k-1 is used as a training sample.
[0101] In some embodiments, training a neural network based on training samples to obtain an initial model; validating the initial model based on validation samples to obtain a first model includes: normalizing the training sample data and the validation sample data; discretizing the training sample data and the validation sample data; training the neural network based on the processed training sample data to obtain an initial model; validating the initial model based on the processed validation sample data to obtain a first model. Normalizing the data, the processed data S ij The calculation formula is as follows:
[0102]
[0103] Among them, x ijis the sample before normalization, S ij is the normalized sample, min(x j ) is the minimum value of the original sample, max(x j ) is the original sample maximum value.
[0104] Discretize the data model, and the processed data z ij The calculation formula is as follows:
[0105]
[0106] Among them, z ij is the discretized sample, min(s j ) is the minimum value of the sample after discretization, max(s j ) is the maximum value of the sample after discretization, Q is the step size, and the calculation formula is as follows:
[0107]
[0108] In one example, the processed result is shown in Table 3, where each column is a feature data, and each row includes the total number (count), mean (mean), standard deviation (std), minimum value (min), 25th percentile (25%), 50th percentile (50%), 75th percentile (75) and maximum value (max):
[0109] Table 3
[0110]
[0111] The product determination method provided in the embodiment of the present application can further improve the accuracy of the product scores output by the model through grouping, normalization and discretization of product data, thereby making the accuracy of determining the exited products higher.
[0112] In some embodiments, the neural network is trained based on the processed training sample data, and before the initial model is obtained, the method further includes: performing null value processing on the sample data and the verification data based on the average value filling method. The null values of the sample data and the verification data are filled using the average value filling method. For example, the filling result is shown in Table 4. When the number of products is 473, there is no null value in each feature data after filling, such as ASP_ID:
[0113] Table 4
[0114]
[0115]
[0116] The product determination method provided in the embodiment of the present application can effectively reduce errors and improve the accuracy of determining the exited products by filling in empty values.
[0117] In some embodiments, during the training of the model, the accuracy of the credit card product exit sign prediction model is tested from three aspects: detailed accuracy, confusion matrix, and node error rate. The grid search (GridSearchCV) method is used to adjust the model parameters for the best parameters. The adjustment results are shown in Table 5:
[0118] Table 5
[0119] parameter Parameter Value learning_rate 0.5 max_depth 2 n_estimators 100 min_child_weight 5 seed 0 subsample 0.5 colsample_bytree 1 gamma 0.04 reg_alpha 0.1 reg_lambda 1
[0120] Among them, the detailed accuracy includes: true positive rate (TP rate), false positive rate (FP rate), precision (Precision), recall rate (Recall), recall rate and harmonic mean of recall rate (F-measure).
[0121] In one embodiment, the output results of the exit model without weight change are shown in Table 6. The target score of the product No. 00000336 is 0.982, the target score of the product No. 00000281 is 0.973, and the target score of the product No. 00000359 is 0.961:
[0122] Table 6
[0123] Product Number Score 00000336 0.982 00000281 0.973 00000359 0.961 … …
[0124] The output results of the weighted exit model are shown in Table 7. The target score of product No. 00000336 is 89.12151, and the target score of product No. 00000281 is 34.18247:
[0125] Table 7
[0126] Product Number Score 00000336 89.12151 00000281 34.18247 … …
[0127] The product determination method provided in the embodiment of the present application can obtain target product data, perform calculations based on the target product data and a trained exit model to obtain a score for the target product, and use target products with scores greater than a first preset threshold as exit products, thereby accurately determining products that need to be exited in a short time.
[0128] Figure 2 Schematic diagram of a product determination device provided in an embodiment of the present application. Figure 2 As shown, the device 200 may include an acquisition module 210 , a calculation module 220 and a determination module 230 .
[0129] The acquisition module 210 is used to acquire target product data, where the target product data includes a target product feature type and product feature data corresponding to the target product feature type.
[0130] Calculation module 220 is used to calculate the target score of the target product based on the exit model and the target product data. The exit model is trained by the product data. The product data includes product feature type, data corresponding to the product feature data type and product exit data. The exit model calculates the target score based on the product feature data corresponding to the target product feature type.
[0131] The determination module 230 is used to determine that the target product is an exit product when the target score of the target product is greater than a first preset threshold.
[0132] The product determination device provided in the embodiment of the present application can calculate the target product score based on the target product data and the exit model, compare the obtained target product score with the first preset threshold, and determine the exit product, thereby improving the accuracy of the exit product and reducing the time spent in determining the exit product.
[0133] In some embodiments, the device also includes: an acquisition module 210, which is also used to acquire product data; a training module 240, which is used to train a neural network based on the product data to obtain a first model; an input module 250, which is used to input the product data into the first model to obtain an importance score of a product feature type; a selection module 260, which is used to select a product feature type whose importance score is greater than a preset threshold as a target feature type; a screening module 270, which is used to screen the product data to obtain product feature data corresponding to the target feature type; the training module 240 is also used to train the first model based on the product feature data corresponding to the target feature and the product exit data to obtain an exit model.
[0134] The product determination device provided in the embodiment of the present application can use the product data of the determined exit data to train the model, screen the data according to the importance of product features calculated by the model, and further train the model using features with higher importance, so that the product scores output by the model are more accurate, and the accuracy of determining the exited products is higher.
[0135] In some embodiments, the training module is further used to train the first model based on the product feature data and product exit data corresponding to the target feature to obtain the exit model, including: the determination module is further used to determine the target weight coefficient of the target feature according to the variable weight formula, the variable weight formula is
[0136]
[0137] Among them, w i is the weight coefficient of the target feature in the exit model, is the target weight coefficient of the target feature, n is the number of target features, i=1, 2, ..., n; the training module is also used to calculate the target product feature data based on the target weight coefficient of the target feature to obtain the target score of the target product.
[0138] In some embodiments, the training module 240 is used to train the neural network based on the product data to obtain the first model, including: a division module 280, used to evenly divide the product data into k groups, wherein group 1 is used as a test sample and k-1 groups are used as training samples, wherein k is greater than or equal to 2; the training module 240 is used to train the neural network based on the training samples to obtain the initial model; and the verification module 290 is used to verify the initial model based on the verification samples to obtain the first model.
[0139] In some embodiments, the training module 240 is used to train the neural network based on the training samples to obtain the initial model; the verification module 290 is used to verify the initial model based on the verification samples to obtain the first model, including: a processing module 201, used to normalize the training sample data and the verification sample data; the processing module 201 is also used to discretize the training sample data and the verification sample data; the training module 240 is used to train the neural network based on the processed training sample data to obtain the initial model; the verification module 290 is used to verify the initial model based on the processed verification sample data to obtain the first model.
[0140] In some embodiments, the product feature type includes at least one of performance, customer base, market, management attributes, risk data, and product attribute data.
[0141] The product determination device provided in the embodiment of the present application can further improve the accuracy of the product scores output by the model through grouping, normalization and discretization of product data, thereby making the accuracy of determining the exited products higher.
[0142] In some embodiments, the apparatus further includes: a processing module 201, further configured to perform null value processing on the sample data and verification data based on a mean value filling method.
[0143] The product determination device provided in the embodiment of the present application can obtain target product data, perform calculations based on the target product data and a trained exit model to obtain a score for the target product, and use target products with scores greater than a first preset threshold as exit products, thereby accurately determining products that need to be exited in a short time.
[0144] Figure 2 Each module in the device shown has the function of realizing Figure 1The product determines the functions of each step performed by the device and can achieve its corresponding technical effects. For the sake of concise description, it will not be repeated here.
[0145] Figure 3 A schematic diagram of the hardware structure of the product provided in the embodiment of the present application is shown.
[0146] The product determination device may include a processor 301 and a memory 302 storing computer program instructions.
[0147] Specifically, the processor 301 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.
[0148] The memory 302 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 302 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In one example, the memory 302 may include a removable or non-removable (or fixed) medium, or the memory 302 is a non-volatile solid-state memory. The memory 302 may be inside or outside the integrated gateway disaster recovery device.
[0149] In one example, the memory 302 may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Thus, typically, the memory 302 includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present application.
[0150] The processor 301 reads and executes the computer program instructions stored in the memory 302 to implement Figure 1 The method / steps S110 to S130 in the embodiment shown in the figure are achieved Figure 1 The corresponding technical effects achieved by executing the methods / steps in the example shown are not repeated here for the sake of brevity.
[0151] In one example, the product determination device may further include a communication interface 303 and a bus 310. Figure 3 As shown, the processor 301, the memory 302, and the communication interface 303 are connected via a bus 310 and communicate with each other.
[0152] The communication interface 303 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.
[0153] Bus 310 includes hardware, software or both, and the parts of product determination equipment are coupled to each other.For example, but not limitation, bus may include accelerated graphics port (Accelerated Graphics Port, AGP) or other graphics bus, enhanced industry standard architecture (Extended Industry Standard Architecture, EISA) bus, front side bus (Front Side Bus, FSB), hypertransmission (Hyper Transport, HT) interconnection, industry standard architecture (IndustryStandard Architecture, ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 310 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the application considers any suitable bus or interconnection.
[0154] The product determination device can execute the product determination method in the embodiment of the present application based on the product data and the exit model, thereby realizing the combination Figure 1 Describes the product determination method.
[0155] In addition, in combination with the product determination method in the above embodiment, the embodiment of the present application can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any one of the product determination methods in the above embodiment is implemented.
[0156] An embodiment of the present application provides a computer program product. When instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the above-mentioned product determination method.
[0157] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.
[0158] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), appropriate firmware, plug-in, function card, etc. When implemented in software, the elements of the present application are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or communication link by a data signal carried in a carrier. "Machine-readable medium" may include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (Radio Frequency, RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.
[0159] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.
[0160] The above reference is according to the method of the embodiment of the present application, the flow chart of the device (system) and the computer program product and / or the block diagram described various aspects of the present application.It should be understood that each square box in the flow chart and / or the block diagram and the combination of each square box in the flow chart and / or the block diagram can be realized by computer program instructions.These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the realization of the function / action specified in one or more square boxes of the flow chart and / or the block diagram.Such a processor can be but is not limited to a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit.It can also be understood that each square box in the block diagram and / or the flow chart and the combination of the square boxes in the block diagram and / or the flow chart can also be realized by the dedicated hardware that performs the specified function or action, or can be realized by the combination of dedicated hardware and computer instructions.
[0161] The above is only a specific implementation of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.
Claims
1. A product determination method, characterized in that: include: Acquire target product data, wherein the target product data includes a target product feature type and product feature data corresponding to the target product feature type; Calculate the target score of the target product based on the exit model and the target product data, the exit model is trained by product data, the product data includes product feature type, data corresponding to the product feature data type and product exit data, and the exit model calculates the target score according to the product feature data corresponding to the target product feature type; When the target score of the target product is greater than a first preset threshold, determining the target product as an exit product; Before calculating the target score of the target product based on the exit model and the target product data, the method further includes: Get product data; Training a neural network based on the product data to obtain a first model; Inputting the product data into the first model to obtain the importance score of the product feature type; Selecting the product feature type whose importance score is greater than a preset threshold as the target feature type; Filtering the product data to obtain product feature data corresponding to the target feature type; Training the first model based on the product feature data and product exit data corresponding to the target feature type to obtain the exit model; The step of training the first model based on the product feature data and product exit data corresponding to the target feature type to obtain the exit model includes: Normalizing and discretizing the product feature data corresponding to the target feature type to obtain processed product feature data corresponding to the target feature type; The first model is trained based on the product feature data and product exit data corresponding to the processed target feature type to obtain the exit model, and the exit model is adjusted based on detailed accuracy, confusion matrix and node error rate; Among them, the normalization process meets the following conditions: Among them, x ij is the product feature data before normalization, S ij is the normalized product feature data, min(x j ) is the minimum value in the product characteristic data, max(x j ) is the maximum value in the product feature data; The discretization process meets the following conditions: Among them, z ij is the discretized product feature data, min(s j ) is the minimum value of the discretized product feature data, max(s j ) is the maximum value of the discretized product feature data, Q is the step length, and the calculation formula is as follows:
2. The method according to claim 1, characterized in that The step of training the first model based on the product feature data and product exit data corresponding to the target feature type to obtain the exit model includes: The target weight coefficient of the target feature type is determined according to the variable weight formula, and the variable weight formula is: in, is the target weight coefficient of the i-th target feature type, w i is the weight coefficient of the i-th target feature type in the exit model, x i is the evaluation value of the i-th target feature type, n is the number of target feature types, i = 1, 2, ..., n; Based on the target weight coefficient of the target feature type, the target product feature data is calculated to obtain the target score of the target product.
3. The method according to claim 1 or 2, characterized in that: The step of training a neural network based on the product data to obtain a first model includes: The product data is evenly divided into k groups, wherein 1 group is used as a verification sample and k-1 groups are used as training samples, where k is greater than or equal to 2; Training the neural network based on the training samples to obtain an initial model; The initial model is verified based on the verification sample to obtain a first model.
4. The method according to claim 3, characterized in that The neural network is trained based on the training samples to obtain an initial model; The initial model is verified based on the verification sample to obtain a first model, including: Normalizing the training sample data and the verification sample data; Performing discretization processing on the training sample data and the verification sample data; The neural network is trained based on the processed training sample data to obtain an initial model; The initial model is verified based on the processed verification sample data to obtain a first model.
5. The method according to claim 1, characterized in that The product feature type includes at least one of performance, customer base, market, management attributes, risk data and product attribute data.
6. The method according to claim 4, characterized in that Before training the neural network based on the processed training sample data to obtain the initial model, the method further includes: Null value processing is performed on the sample data and verification data based on the mean value filling method.
7. A product determination device, characterized in that: The device comprises: An acquisition module, used to acquire target product data, wherein the target product data includes a target product feature type and product feature data corresponding to the target product feature type; a calculation module, configured to calculate a target score for the target product based on an exit model and the target product data, wherein the exit model is trained by product data, the product data includes a product feature type, data corresponding to the product feature data type, and product exit data, and the exit model calculates the target score based on the product feature data corresponding to the target product feature type; A determination module, configured to determine that the target product is an exit product when the target score of the target product is greater than a first preset threshold; The device also includes: The acquisition module is also used to acquire product data; A training module, used for training a neural network based on the product data to obtain a first model; An input module, used for inputting the product data into the first model to obtain the importance score of the product feature type; A selection module, used for selecting a product feature type whose importance score is greater than a preset threshold as a target feature type; A screening module, used for screening the product data to obtain product feature data corresponding to the target feature type; The training module is further used to train the first model based on the product feature data and product exit data corresponding to the target feature type to obtain the exit model; The training module is specifically used for: Normalizing and discretizing the product feature data corresponding to the target feature type to obtain processed product feature data corresponding to the target feature type; The first model is trained based on the product feature data and product exit data corresponding to the processed target feature type to obtain the exit model, and the exit model is adjusted based on detailed accuracy, confusion matrix and node error rate; Among them, the normalization process meets the following conditions: Among them, x ij is the product feature data before normalization, S ij is the normalized product feature data, min(x j ) is the minimum value in the product characteristic data, max(x j ) is the maximum value in the product feature data; The discretization process meets the following conditions: Among them, z ij is the discretized product feature data, min(s j ) is the minimum value of the discretized product feature data, max(s j ) is the maximum value of the discretized product feature data, Q is the step length, and the calculation formula is as follows:
8. The device according to claim 7, characterized in that The training module is further used to train the first model based on the product feature data and product exit data corresponding to the target feature type to obtain the exit model, including: The determination module is also used to determine the target weight coefficient of the target feature type according to the variable weight formula, and the variable weight formula is in, is the target weight coefficient of the i-th target feature type, w i is the weight coefficient of the i-th target feature type in the exit model, x i is the evaluation value of the i-th target feature type, n is the number of target feature types, i = 1, 2, ..., n; The training module is also used to calculate the target product feature data based on the target weight coefficient of the target feature type to obtain the target score of the target product.
9. The device according to claim 7 or 8, characterized in that The training module is used to train a neural network based on the product data to obtain a first model, including: A division module, used to evenly divide the product data into k groups, wherein 1 group is used as a verification sample and k-1 groups are used as training samples, wherein k is greater than or equal to 2; The training module is used to train the neural network based on the training samples to obtain an initial model; The verification module is used to verify the initial model based on the verification sample to obtain a first model.
10. The device according to claim 9, characterized in that The training module is used to train the neural network based on the training samples to obtain an initial model; The verification module is used to verify the initial model based on the verification sample to obtain a first model, including: A processing module, used for normalizing the training sample data and the verification sample data; The processing module is further used to discretize the training sample data and the verification sample data; The training module is used to train the neural network based on the processed training sample data to obtain an initial model; The verification module is used to verify the initial model based on the processed verification sample data to obtain a first model.
11. The device according to claim 7, characterized in that The product feature type includes at least one of performance, customer base, market, management attributes, risk data and product attribute data.
12. The device according to claim 10, characterized in that The device also includes: The processing module is also used to perform null value processing on the sample data and verification data based on the mean value filling method.
13. A product determination device, characterized in that: The product determination device includes: a processor, and a memory storing computer program instructions; The processor reads and executes the computer program instructions to implement the product determination method according to any one of claims 1-6.
14. A computer storage medium, characterized in that: The computer storage medium stores computer program instructions, and when the computer program instructions are executed by the processor, the product determination method according to any one of claims 1 to 6 is implemented.
15. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the product determination method as described in any one of claims 1 to 6.
Citation Information
Patent Citations
Operating state assessment method of capacitor voltage transformer
CN103576050A
Network attack risk assessment method of distribution network CPS dependent system based on attacker perspective
CN109191326A
Product pushing method based on neural network and related equipment thereof
CN110458637A
User credit score obtaining method and device, server and storage medium
CN110544155A
Characteristic importance sorting method and device, equipment and storage medium
CN110738527A
