CN117035472A - Enterprise operation health assessment method, model construction method and device - Google Patents
Enterprise operation health assessment method, model construction method and device Download PDFInfo
- Publication number
- CN117035472A CN117035472A CN202210460085.6A CN202210460085A CN117035472A CN 117035472 A CN117035472 A CN 117035472A CN 202210460085 A CN202210460085 A CN 202210460085A CN 117035472 A CN117035472 A CN 117035472A
- Authority
- CN
- China
- Prior art keywords
- enterprise
- features
- model
- feature
- quantitative evaluation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000036541 health Effects 0.000 title claims abstract description 93
- 238000000034 method Methods 0.000 title claims abstract description 73
- 238000010276 construction Methods 0.000 title claims abstract description 17
- 238000011158 quantitative evaluation Methods 0.000 claims abstract description 60
- 238000012549 training Methods 0.000 claims abstract description 60
- 238000012545 processing Methods 0.000 claims abstract description 58
- 238000011156 evaluation Methods 0.000 claims abstract description 24
- 230000009466 transformation Effects 0.000 claims abstract description 23
- 238000012360 testing method Methods 0.000 claims abstract description 11
- 238000012216 screening Methods 0.000 claims description 36
- 238000004590 computer program Methods 0.000 claims description 16
- 238000007781 pre-processing Methods 0.000 claims description 14
- 238000013139 quantization Methods 0.000 claims description 7
- 238000012106 screening analysis Methods 0.000 claims description 5
- 238000003860 storage Methods 0.000 claims description 5
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 description 11
- 238000003066 decision tree Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 230000005540 biological transmission Effects 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 238000007477 logistic regression Methods 0.000 description 5
- 238000011985 exploratory data analysis Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000001276 Kolmogorov–Smirnov test Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 239000000654 additive Substances 0.000 description 2
- 230000000996 additive effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000013210 evaluation model Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012797 qualification Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 108010001267 Protein Subunits Proteins 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000007717 exclusion Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000013488 ordinary least square regression Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Educational Administration (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Development Economics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to an enterprise operation health assessment method, a model construction method and a device, wherein the model construction method comprises the following steps: processing the original basic data of the enterprise to obtain multi-dimensional characteristics; sample processing is carried out on the multidimensional features, and high-dimensional features of the renewal enterprise samples and the loss enterprise samples are selected; performing feature transformation selection on the high-dimensional features to determine an optimal feature combination; dividing the optimal feature combination into a training set and a testing set, taking a renewal enterprise and a loss enterprise as two classification targets for supervised model training and learning, and combining preset sectional renewal rates to construct a quantitative evaluation model. The application solves the technical problems of difficulty in quantitative evaluation or low evaluation accuracy, lower result stability or poor interpretation of the enterprise operation health degree, realizes the accurate stability and the robust performance of the quantitative evaluation result of the enterprise operation health degree, and remarkably improves the identification efficiency and the overall continuous rate of the enterprise operation condition.
Description
Technical Field
The application relates to the technical field of enterprise evaluation, in particular to an enterprise operation health evaluation method, a model construction method and a model construction device.
Background
In the field of providing services to an enterprise (to B), it is necessary to provide insight into the status and health of service enterprise operations, so as to perform more refined and personalized service operation actions, so as to improve the experience viscosity and operation efficiency of the service enterprise. Especially in the field of SaaS enterprise service, the continuous rate is an extremely critical index, and the health degree of service enterprise operation is quantitatively evaluated and accurately predicted, so that the overall continuous rate of the to B enterprise is directly determined.
Currently, business health assessment is roughly classified into two types.
One is a business experience scoring based scheme that uses commodity transaction totals (Gross Merchandise Volume, GMV) operated by an enterprise as an indicator of health. The method has the defects that whether the business operation condition is healthy or not is influenced by a plurality of factors, the judgment is carried out according to a single index dimension, and the accuracy of evaluating the health degree of the business operation is low.
In addition, the method is a scheme based on an unsupervised learning model, wherein the scheme is to construct clustering algorithms such as K-means and DBscan through different dimensions, perform training on original basic data of enterprises in an unsupervised mode of clustering by people so as to divide the enterprise groups with health degree levels. The scheme has the defects that the operation health condition of an enterprise cannot be comprehensively and quantitatively evaluated, only clustering and grouping can be performed, and quantitative evaluation results cannot be output.
Aiming at the problems that in the related art, based on an unsupervised learning model, training is carried out based on original basic data of enterprises, clustering and grouping can be carried out only, quantized evaluation results cannot be output, and the accuracy of the evaluation results is low, no effective solution is proposed at present.
Disclosure of Invention
In this embodiment, an enterprise operation health assessment method, a model construction method and an enterprise operation health assessment device are provided, so as to solve the problem that in the related art, training is performed based on original basic data of an enterprise, clustering and grouping can only be performed, quantized assessment results cannot be output, and the accuracy of the assessment results is low.
In a first aspect, in this embodiment, there is provided a model building method, including:
processing the original basic data of the enterprise to obtain multi-dimensional characteristics;
sample processing is carried out on the multidimensional features, and high-dimensional features of renewal enterprise samples and loss enterprise samples are selected;
performing feature transformation selection on the high-dimensional features to determine an optimal feature combination;
dividing the optimal feature combination into a training set and a testing set, taking the renewal enterprise and the loss enterprise as two classification target variables of training and learning, and training and learning the original network model with supervised learning by combining with each preset segmentation renewal rate to construct a quantitative evaluation model.
In some embodiments, the processing the sample of the multidimensional feature to select a high-dimensional feature of a renewal enterprise sample and a loss enterprise sample includes:
preprocessing the multi-dimensional characteristics according to a preset preprocessing rule;
and carrying out sample processing on the preprocessed multidimensional features, and selecting the high-dimensional features of the pay-out enterprise samples and the loss enterprise samples.
In some embodiments, the sample processing the preprocessed multidimensional feature, selecting the high-dimensional feature of the pay-out enterprise sample and the churn enterprise sample includes:
sample processing is carried out on the preprocessed multidimensional features according to a preset screening strategy, and high-dimension features of the output renewal enterprise samples and the loss enterprise samples are selected;
the screening strategy comprises a service life threshold and an enterprise type sample division principle.
In some of these embodiments, the feature transformation selecting the high-dimensional feature to determine an optimal feature combination includes:
performing feature primary screening on the high-dimensional features to obtain primary screening features;
and optimally selecting the primary screening features by using a feature use model to obtain an optimal feature combination.
In some embodiments, the performing feature prescreening on the high-dimensional feature to obtain a prescreened feature includes:
grouping the high-dimensional features by using a chi-square box, and determining a first grouping feature and a corresponding optimal grouping threshold;
converting the optimal grouping threshold value of each group of the first grouping features by using evidence weight coding, and selecting a second grouping feature from the first grouping features based on a preset information value;
and performing multi-feature correlation screening analysis and stability evaluation on the second grouping feature to obtain a primary screening feature.
In some of these embodiments, the method further comprises:
and determining an expression of the health and the prediction probability based on the influence weight of each feature output by the quantitative evaluation model on the operation health.
In some embodiments, the health and predictive probability is expressed as:
wherein Score represents a health Score; p represents a prediction probability; a and B represent constants.
In a second aspect, in this embodiment, there is provided an enterprise operation health assessment method, including:
carrying out quantitative evaluation on the original basic data of the enterprise to be analyzed by using the quantitative evaluation model constructed by the model construction method according to any one of the first aspects to obtain a quantitative evaluation result of the enterprise to be analyzed.
In some of these embodiments, the method further comprises:
and inputting the quantitative evaluation result into an expression of health and prediction probability to obtain the business health score and the health degree of the enterprise to be analyzed.
In a third aspect, in this embodiment, there is provided a model building apparatus including: the device comprises a processing module, a sample processing module, a characteristic transformation selection module and a training learning module;
the processing module is used for processing the original basic data of the enterprise to obtain multidimensional features;
the sample processing module is used for processing the samples of the multidimensional features and selecting high-dimensional features of the renewal enterprise samples and the loss enterprise samples;
the feature transformation selection module is used for carrying out feature transformation selection on the high-dimensional features so as to determine an optimal feature combination;
the training learning module is used for dividing the optimal feature combination into a training set and a testing set, taking the renewal enterprise and the loss enterprise as two kinds of target variables of training learning, and combining preset sectional charge rates to perform training learning on the original network model with supervised learning so as to construct a quantitative evaluation model.
In a fourth aspect, in this embodiment, there is provided an enterprise operation health assessment apparatus, including: a quantization evaluation module;
the quantitative evaluation module is configured to perform quantitative evaluation on the original basic data of the enterprise to be analyzed by using the quantitative evaluation model constructed by the model construction method according to any one of the first aspects, so as to obtain a quantitative evaluation result of the enterprise to be analyzed.
In a fifth aspect, in this embodiment, there is provided a computer device including a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the model building method according to the first aspect, when the processor executes the computer program; or, implementing the enterprise operation health assessment method according to the second aspect.
In a sixth aspect, in this embodiment, there is provided a storage medium having stored thereon a computer program which, when executed by a processor, implements the model building method of the first aspect described above; or, implementing the enterprise operation health assessment method according to the second aspect.
Compared with the related art, the enterprise operation health assessment method, the model construction method and the device provided in the embodiment. The method for constructing the model comprises the steps of processing original basic data of an enterprise to obtain multidimensional features; sample processing is carried out on the multidimensional features, and high-dimensional features of the renewal enterprise samples and the loss enterprise samples are selected; performing feature transformation selection on the high-dimensional features to determine an optimal feature combination; the method comprises the steps of dividing an optimal feature combination into a training set and a testing set, taking a renewal enterprise and a loss enterprise as two kinds of target variables of training and learning, and training and learning an original network model with supervision and learning by combining preset sectional continuous rates to construct a quantitative evaluation model, so that the problems that the original basic data of the enterprise is trained, only clustering and grouping can be carried out, quantized evaluation results cannot be output, the accuracy of the evaluation results is low, the output of the quantized evaluation results is realized, the accuracy, stability and the robustness of the enterprise operation health quantitative evaluation results can be guaranteed, and the identification efficiency and the overall continuous rate of the enterprise operation condition are remarkably improved.
The details of one or more embodiments of the application are set forth in the accompanying drawings and the description below to provide a more thorough understanding of the other features, objects, and advantages of the application.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute a limitation on the application. In the drawings:
fig. 1 is a hardware block diagram of a terminal device of a model building method according to an embodiment of the present application;
FIG. 2 is a flow chart of a model building method according to an embodiment of the present application;
fig. 3 is a flowchart of step S230 in fig. 2;
FIG. 4 is a block diagram of a model building apparatus according to an embodiment of the present application;
FIG. 5 is a flowchart of an enterprise business health assessment method according to an embodiment of the present application;
FIG. 6 is a flow chart of an enterprise business health assessment method according to a preferred embodiment of the present application;
fig. 7 is a block diagram of an enterprise operation health assessment apparatus according to an embodiment of the present application.
In the figure: 210. a processing module; 220. a sample processing module; 230. a feature transformation selection module; 240. training a learning module; 510. and a quantitative evaluation module.
Detailed Description
The present application will be described and illustrated with reference to the accompanying drawings and examples for a clearer understanding of the objects, technical solutions and advantages of the present application.
Unless defined otherwise, technical or scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terms "a," "an," "the," "these" and similar terms in this application are not intended to be limiting in number, but may be singular or plural. The terms "comprising," "including," "having," and any variations thereof, as used herein, are intended to encompass non-exclusive inclusion; for example, a process, method, and system, article, or apparatus that comprises a list of steps or modules (units) is not limited to the list of steps or modules (units), but may include other steps or modules (units) not listed or inherent to such process, method, article, or apparatus. The terms "connected," "coupled," and the like in this disclosure are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The term "plurality" as used herein means two or more. "and/or" describes an association relationship of an association object, meaning that there may be three relationships, e.g., "a and/or B" may mean: a exists alone, A and B exist together, and B exists alone. Typically, the character "/" indicates that the associated object is an "or" relationship. The terms "first," "second," "third," and the like, as referred to in this disclosure, merely distinguish similar objects and do not represent a particular ordering for objects.
The enterprise is enterprise-oriented (to Business, to B): a business model provides services to businesses, which may also be merchants.
Commodity transaction total (Gross Merchandise Volume, GMV): is the meaning of the sum of the deals (within a certain period of time). Commonly used in the e-commerce industry, typically involves placing an unpaid order amount.
Evidence weight (Weight of Evidence, WOE): the method is a supervised coding mode, takes the attribute of the concentration degree of the prediction category as the numerical advantage of coding, and normalizes the value of the characteristic to a similar scale.
Exploratory data analysis (Exploratory Data Analysis, EDA): the method is a data analysis method for exploring the structure and rule of the existing data (especially the original data obtained by investigation or observation) under the prior assumption as few as possible by means of drawing, tabulation, equation fitting, characteristic quantity calculation and the like.
Gradient-lifted decision tree (Gradient Boosting Decision Tree, GBDT): is an iterative decision tree algorithm, which consists of a plurality of decision trees, and the conclusions of all the trees are accumulated as the final answer.
Gradient lifting lightweight framework (light gradient boosting machine, lightGBM) based on decision tree: is a framework for realizing the idea of GBDT algorithm, and is a boosting decision tree tool which is opened by Microsoft DMTC team.
Logistic regression (Logistic Regression, LR): is a machine learning method for solving the problem of two classifications (0 or 1) for estimating the likelihood of something. Such as the likelihood of a renewal of a business, the likelihood of a patient suffering from a disease, the likelihood of an advertisement being clicked on by a user, etc.
Software-as-a-Service (SaaS): is an application mode for providing software services based on the internet. The SaaS is based on IaaS (Infrastructure as a Service ) and PaaS (Platform as a Service, platform as a service), and is the uppermost layer of the cloud service and one layer of the direct users.
Information value (Information Value, IV): i.e. the information value index, is a common index in the scoring card model.
Coefficient of variance expansion (variance inflation factor, VIF): is a measure of the severity of complex (multiple) co-linearity in a multiple linear regression model. It represents the ratio of the variance of the regression coefficient estimator compared to the variance if the nonlinear correlation between the independent variables is assumed.
Population stability index (Population Stability Index, PSI): is an index for measuring the deviation between the predicted value and the actual value of the model.
Saprolil value additive interpretation (SHapley Additive explanation, SHAP): a predictive value is generated for each sample model, shape value being the value assigned to each feature in the sample.
Red pool information content criterion (Akaike information criterion, AIC): a criterion for measuring the Goodness of fit (Goodness of fit) of a statistical model.
Bayesian information criteria (Bayesian InformationCriterion, BIC): bayesian information criteria are similar to AIC for model selection.
Common least squares method (Ordinary Least Squares, OLS): the fitting criterion is given, i.e. the sum of squares of the differences between the estimated values and the actual observed values of the interpreted variables should be minimized.
The method embodiments provided in the present embodiment may be executed in a terminal, a computer, or similar computing device. For example, the method runs on a terminal, and fig. 1 is a block diagram of the hardware structure of the terminal of the model building method of the present embodiment. As shown in fig. 1, the terminal may include one or more (only one is shown in fig. 1) processors 102 and a memory 104 for storing data, wherein the processors 102 may include, but are not limited to, a microprocessor MCU, a programmable logic device FPGA, or the like. The terminal may also include a transmission device 106 for communication functions and an input-output device 108. It will be appreciated by those skilled in the art that the structure shown in fig. 1 is merely illustrative and is not intended to limit the structure of the terminal. For example, the terminal may also include more or fewer components than shown in fig. 1, or have a different configuration than shown in fig. 1.
The memory 104 may be used to store a computer program, for example, a software program of application software and a module, such as a computer program corresponding to a model building method in the present embodiment, and the processor 102 executes the computer program stored in the memory 104 to perform various functional applications and data processing, that is, to implement the above-described method. Memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include memory remotely located relative to the processor 102, which may be connected to the terminal via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The transmission device 106 is used to receive or transmit data via a network. The network includes a wireless network provided by a communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, simply referred to as NIC) that can connect to other network devices through a base station to communicate with the internet. In one example, the transmission device 106 may be a Radio Frequency (RF) module, which is configured to communicate with the internet wirelessly. In other embodiments, the enterprise operation health assessment method may also be executed in the above-mentioned computing device.
In this embodiment, a method for constructing a model is provided, fig. 2 is a flowchart of the method for constructing a model of this embodiment, and as shown in fig. 2, the flowchart includes the following steps:
step S210, processing original basic data of an enterprise to obtain multidimensional features;
the enterprise refers To an object served in To B, and may also be a merchant. The original basic data of the enterprise refers to relevant original basic data which can be directly acquired, including, but not limited to, operation records, operation flowing water, qualification information, lawsuits, performance records, legal qualification and the like. Such as: in the SaaS service field, the original base number may be obtained from the SaaS system. And are not illustrated for the other fields.
And processing the original basic data according to the dimensions divided in advance to obtain multi-dimensional characteristics. Specific multi-dimensional features include, but are not limited to, operation records, application plugins, after-market services, pneumatic handling, user marketing, and the like. And each dimension feature can be further dimension processed according to the equal dimension of approximately 7 days, 30 days, 60 days, 90 days and the like and the time interval dimension. In other embodiments, the above-mentioned raw base data may also be subjected to derivative processing, so as to expand the dimension of the sample and improve the accuracy of model training. Such as: the operation records can be derived and processed into dimension characteristics such as operation use, active login, transaction and the like; the application range of the model can be improved.
Step S220, sample processing is carried out on the multidimensional features, and high-dimensional features of the renewal enterprise samples and the loss enterprise samples are selected;
sample processing refers to screening of samples, and high-dimension characteristics of a renewal enterprise sample and high-dimension characteristics of a loss enterprise sample are selected from multidimensional characteristics. The high-dimensional characteristic is obtained by performing sample processing on the multi-dimensional characteristic. That is, the high latitude features of other types of enterprise samples do not participate in the training of subsequent models, avoiding noise interference during model learning. Other types of enterprise samples, including but not limited to uncertain enterprise samples, abnormal enterprise samples, and the like.
Step S230, performing feature transformation selection on the high-dimensional features to determine an optimal feature combination;
it is known that the selection of training samples has a very important influence on the training results of the model, directly resulting in the accuracy of the model training. From so many raw base data, how to determine the optimal feature combination is a very difficult process, and each processing step or data selection is different, which results in different optimal feature combinations, and thus completely different model training results. In this embodiment, feature transformation selection refers to a process of searching an optimal feature combination, and determines an optimal feature combination with distinguishing capability and stability through feature transformation selection so as to remove interference and improve rationality and stability of sample selection.
And step S240, dividing the optimal feature combination into a training set and a testing set, taking a renewal enterprise and a loss enterprise as two kinds of target variables of training learning, and training and learning the original network model with supervised learning by combining preset sectional renewal rates to construct a quantitative evaluation model.
The training set and the test set may be divided into a certain proportion, such as 4:1, 3:1, 5:1, etc. The original network model with supervised learning is: logistic Regression model. And training and learning the Logistic Regression model by taking the renewal enterprise and the loss enterprise as two classification target variables of training and learning and combining preset sectional renewal rates to construct a quantitative evaluation model. The constructed quantitative evaluation model can output the weight of each feature on the influence of the operation health degree, namely the model prediction probability, so that the interpretability of the model is enhanced, and the health degree influence factors can be understood more intuitively. The supervised learning original network model may select other models, such as XGBoost in GBDT decision tree model, random forest RF, RNN of neural network model, etc., without limitation.
In other embodiments, the Area magnitude (AUC) Under the ROC Curve, kolmogorov-Smirnov test (K-S test) values may be used as the effect evaluation index, reducing the overfitting of the training process and improving the predictive power of the model. Where AUC is typically between 0.5 and 1.0, a larger AUC represents better performance. AUC (Area Under roc Curve) is a criterion for measuring the quality of the classification model. Wherein, K-S test: KS is commonly used to evaluate model discrimination as proposed by two math families, a.n.kolmogorov and n.v.smirnov.
Through the steps, samples trained by the model are screened firstly, and the method specifically comprises the following steps: processing the original basic data of the enterprise to obtain multi-dimensional characteristics; sample processing is carried out on the multidimensional features, and high-dimensional features of the renewal enterprise samples and the loss enterprise samples are selected; performing feature transformation selection on the high-dimensional features to determine an optimal feature combination, wherein the optimal feature combination which has distinguishing capability and is relatively stable can be determined; based on the optimal feature combination, taking a renewal enterprise and a loss enterprise as two kinds of target variables of training and learning, and training and learning an original network model with supervised learning by combining preset sectional renewal rates to construct a quantitative evaluation model; the training of the quantitative evaluation model capable of outputting the quantitative evaluation result is realized, the accuracy, stability and robustness of the quantitative evaluation result of the enterprise operation health degree can be ensured, and the identification efficiency and the overall continuous rate of the enterprise operation condition are remarkably improved; the method solves the technical problems that the training is carried out based on the original basic data of enterprises, only clustering and grouping can be carried out, quantitative evaluation or the evaluation accuracy is not high, the result stability is low or the interpretation is poor in the industry.
In some of these embodiments, step S220 includes the steps of:
step S221, preprocessing the multidimensional feature according to a preset preprocessing rule;
step S222, sample processing is carried out on the preprocessed multidimensional features, and high-dimensional features of the renewal enterprise samples and the loss enterprise samples are selected.
And preprocessing the multidimensional features according to a preset preprocessing rule. The method comprises the following steps: EDA exploration is carried out on the multi-dimensional features, including missing values, outlier data preprocessing and the like, multi-dimensional features (row by row) or samples (column by column) with the missing rate exceeding 70% are eliminated, multi-dimensional features and the like in a numerical distribution set with the same value accounting for over 90% are eliminated and the like. By preprocessing the multidimensional features, interference can be reduced, and the integrity of model training can be effectively improved.
In some of these embodiments, step S222 includes the steps of:
sample processing is carried out on the preprocessed multidimensional features according to a preset screening strategy, and high-dimensional features of the renewal enterprise samples and the loss enterprise samples are selected;
the screening policy includes a age threshold and a business type sample partitioning principle.
Specifically, the service life threshold may be about 1 year, about 2 years, about 3 years, about 4 years, etc. The enterprise type sample division principle is GBIE principle, specifically Good: good clients and renewal enterprises refer to renewal enterprises before the SaaS service expires. Bad: bad clients are lost enterprises, namely enterprises which are not renewed for n days after the SaaS service is expired; indeterminate: an uncertain enterprise refers to an enterprise within n days after the renewal of the enterprise is expired, and the enterprise is influenced by various uncertain factors such as forgetting the renewal of the enterprise, short-term organization adjustment and the like, wherein n days are determined by gradually analyzing the renewal rolling rate (accumulated renewal rate). Exclusion: exclusion-type enterprises refer to enterprises that are subject to wind control penalties due to suspected operational anomalies/fraud, etc. during service.
And selecting a historical service enterprise meeting the service life threshold as a sample, and dividing the enterprise sample into four types of renewal enterprise sample, loss enterprise sample, uncertain enterprise sample and abnormal enterprise sample based on the preprocessed multidimensional characteristic according to the GBIE principle. Screening high-dimensional characteristics of the renewal enterprise samples and high-dimensional characteristics of the loss enterprise samples from the four types; determining the high latitude characteristics corresponding to each type; and then selecting two types of high latitude characteristics, namely a renewal enterprise sample and a loss enterprise sample. The two types are used as two kinds of target variables for model training learning, and other types do not participate in model training, so that noise interference in the model learning process is avoided.
In some of these embodiments, as shown in fig. 3, step S230 includes the steps of:
step S231, performing feature primary screening on the high-dimensional features to obtain primary screening features;
and S232, optimally selecting the primary screening features by using a feature use model to obtain an optimal feature combination.
Specifically, the high-dimensional characteristics of the renewal enterprise sample and the loss enterprise sample are subjected to primary screening, and the primary screening characteristics are optimally selected by utilizing a characteristic use model to obtain an optimal characteristic combination; the effect of feature dimension reduction is achieved, and the learning efficiency of the follow-up model is improved.
Step S231, performing feature prescreening on the high-dimensional features to obtain prescreened features, including the following steps:
grouping the high-dimensional features by using a chi-square box, and determining a first grouping feature and a corresponding optimal grouping threshold;
converting the optimal grouping threshold value of each group of first grouping features by using evidence weight coding, and selecting a second grouping feature from the first grouping features based on a preset information value;
and performing multi-feature correlation screening analysis and stability evaluation on the second grouping feature to obtain a primary screening feature.
Specifically, first, the chi-square bin is used to find the optimal grouping threshold value corresponding to each group of first grouping features. If the data is null, the data is singly in a box, the combination of the optimal sub-boxes is determined by the chi-square value with smaller adjacent intervals, and the chi-square value is calculated as follows:
wherein m is the number of the characteristic values; k is the renewal/churn category number; a is that ij In the i-th group of the characteristics, the observation frequency of k categories; e (E) ij Under the assumption of original A ij Is not limited to the above-described embodiments.
And secondly, screening whether each group of renewal sample rates after the box division has business significance or not, and properly carrying out box division fine adjustment. The business meaning refers to whether the sample continuous rate data of each group shows the same with the actual situation of the business. For example, in business understanding cognition, the more login days of an enterprise are about 1 month, the higher the business health degree and the renewal will, but if the more login days are represented by data, the lower the business health degree and the renewal rate are, the business understanding is not met, the business meaning is not achieved, and the elimination or the fine adjustment of the boxes is needed. The fine tuning of the bin is typically performed only when one or more of the following conditions are met: a: the number of individual sub-boxes is relatively low (less than 5%); b: the data performance of the neighbor sub-boxes (mainly referred to herein as the sub-box renewal enterprise duty ratio) is basically consistent; c: excessive number of bins (more than 10), etc. The number of login days-corresponding bin WOE codes and corresponding scores in the high-dimensional features of approximately 7 days as shown in table 1; as shown in table 1, the binning fine tuning refers to performing a combination or splitting of neighbor bins, such as [1,2], [2,3], fine tuning the combination to [1,3]; splitting [1,3] into [1,2], [2,3], etc.
Because the optimal grouping threshold after the binning is of the type, the WOE is input as a model after being coded and converted, and the WOE is calculated as follows:
wherein P is y1 The proportion of the renewal enterprise samples to all renewal enterprise samples is calculated; p (P) y0 The proportion of the lost enterprise samples to all lost enterprise samples is calculated; b is the total number of the samples of the renewal enterprise; b (B) i The number of renewal samples corresponding to the variable i is set; g is the total number of good samples; g i The number of the loss samples corresponding to the variable i.
Selecting features with higher information value (IV value) (generally IV value is more than or equal to 0.02), namely relatively strong prediction capability, performing model-entering training, and selecting second grouping features; IV values were calculated as follows:
wherein i, n is the number of samples; b is the total number of the samples of the renewal enterprise; b (B) i The number of renewal samples corresponding to the variable i is set; g is the total number of good samples, G i The number of the loss samples corresponding to the variable i.
TABLE 1
And thirdly, carrying out multi-feature correlation screening analysis on the second grouping features, checking the pairwise linear correlation through the Pearson coefficient, measuring the multi-collinearity through a variance expansion factor (VIF), eliminating redundant features, and avoiding the interference to model parameters.
Wherein, the expression of the variance expansion factor is:
wherein R is i And carrying out regression analysis on the other characteristics for the characteristics to obtain negative correlation coefficients.
Meanwhile, considering stability, dividing samples according to different renewal months as an actual reference set and an expected set, calculating PSI value to evaluate characteristic stability, and selecting characteristics which do not have larger fluctuation along with time, thereby completing primary screening characteristics.
Wherein, the PSI value has the following expression:
wherein n represents the feature division n intervals; i is the i-th segment interval;representing the number ratio of the ith value segment of the feature in the actual reference set; />Representing the number of segments of the feature in the expected set that are the ith value.
Specifically, step S232 includes the following steps:
the primary screening features after the primary screening can be trained by combining with the LightGBM model to obtain the feature importance based on the information gain, the influence trend of the important features on enterprise renewal or loss can be interpreted and analyzed according to the SHAP value, the importance of the screening features is higher, the influence interpretation of the SHAP value health degree accords with the features of business significance, and the SHAP value is calculated as follows:
wherein,is the SHAP value for each feature; s is a subset of features used in the LightGBM model; x is a vector of eigenvalues of the samples to be interpreted; p is the number of features; val (S) refers to the model output value under the feature combination S; a total of p features, then these p features share p-! Seed combination, if a certain feature j is fixed, then the remainder is (p-S1) +|! S-! A combination.
Finally, the selected features are subjected to bidirectional stepwise recursion, the ols is used as an estimated value, the BIC is used as a criterion, the model complexity is prevented from being too high, meanwhile, the optimal feature combination for optimal decision finding is learned and found, and the final model entering features are determined, namely, the final model entering features, the number of newly-increased users of the enterprise in the day of about 7 days, the processing time length (hours) of the right maintenance of the enterprise in the day of about 30 days, and the like. The BIC criteria are as follows: bic=kln (n) -2ln (L);
where k is the number of model parameters, n is the number of samples, and L is the likelihood function.
Through the process, the feature selection method of multiple actions is adopted, and through the optimal box division, correlation, stability, importance, interpretability and optimal feature combination of the features, the feature dimension is reduced, the model learning efficiency is improved, and meanwhile, the generalization capability and the overall effect stability of the model are improved.
In some of these embodiments, the model building method further comprises the steps of:
based on the influence weight of each feature output by the quantitative evaluation model on the operation health degree, an expression of health and prediction probability is determined.
The expression of health and prediction probability is:
wherein Score represents a health Score; p represents a prediction probability; a and B represent constants.
The Logistic Regression model predictive probability p can be converted and mapped into a health degree score through the expression of the health and predictive probability, and then the health degree grade of the enterprise is determined through the mapping relation between the operation health and the health degree grade. Table 2 shows the mapping relationship between the operation health and the health level. A and B can be determined by the following conversion formula:
wherein Score 0 Is a benchmark score; PDO is the fraction of the probability doubling; p is model predictive probability; A. b is a constant; ln is a logarithmic function based on a constant e. In one embodiment, the reference score is set to be 45 points, the probability ratio of the probability of the appointed renewal enterprise to the probability of the loss enterprise is 3/8, the probability doubling score PDO is 9 points, and the numerical values of A and B can be obtained. The parameters may also be adapted if enterprise business conditions are quantitatively assessed due to other areas. This is not illustrated one by one.
Table 2:
| health grade | Score range |
| Health care | [80,100] |
| Sub-health | [60,80) |
| Unhealthy | [30,60) |
| Quasi-loss | [0,30) |
The above-described respective modules may be functional modules or program modules, and may be implemented by software or hardware. For modules implemented in hardware, the various modules described above may be located in the same processor; or the above modules may be located in different processors in any combination.
It should be noted that the steps illustrated in the above-described flow or flow diagrams of the figures may be performed in a computer system, such as a set of computer-executable instructions, and that, although a logical order is illustrated in the flow diagrams, in some cases, the steps illustrated or described may be performed in an order other than that illustrated herein.
The embodiment also provides a model building device, which is used for realizing the embodiment and the preferred implementation manner, and is not described in detail. The terms "module," "unit," "sub-unit," and the like as used below may refer to a combination of software and/or hardware that performs a predetermined function. While the means described in the following embodiments are preferably implemented in software, implementations in hardware, or a combination of software and hardware, are also possible and contemplated.
Fig. 4 is a block diagram of the structure of the model building apparatus of the present embodiment, as shown in fig. 4, including: a processing module 210, a sample processing module 220, a feature transformation selection module 230, and a training learning module 240;
a processing module 210, configured to process the original basic data of the enterprise to obtain a multidimensional feature;
The sample processing module 220 is configured to perform sample processing on the multidimensional feature, and select a renewal enterprise sample and a high-dimensional feature of a loss enterprise sample;
a feature transformation selection module 230, configured to perform feature transformation selection on the high-dimensional feature to determine an optimal feature combination;
the training learning module 240 is configured to divide the optimal feature combination into a training set and a testing set, take a renewal enterprise and a loss enterprise as two kinds of target variables of training learning, and perform training learning on the original network model with supervised learning by combining preset segmentation continuous rates to construct a quantized evaluation model.
By the device, the training of the quantitative evaluation model capable of outputting the quantitative evaluation result is realized, the accuracy, stability and robustness of the quantitative evaluation result of the enterprise operation health degree can be ensured, and the identification efficiency and the overall continuous rate of the enterprise operation condition are remarkably improved; the method solves the problems that training is carried out based on original basic data of enterprises, clustering and grouping can be carried out only, quantized evaluation results cannot be output, and the accuracy of the evaluation results is low.
In some of these embodiments, the sample processing module 220 includes a preprocessing unit and a sample processing unit;
The preprocessing unit is used for preprocessing the multidimensional features according to a preset preprocessing rule;
the sample processing unit is used for processing the preprocessed multidimensional features and selecting high-dimensional features of the renewal enterprise samples and the loss enterprise samples.
In some embodiments, the sample processing unit is further configured to perform sample processing on the preprocessed multidimensional feature according to a preset screening policy, and select a renewal enterprise sample and a high-dimensional feature of a loss enterprise sample;
the screening policy includes a age threshold and a business type sample partitioning principle.
In some of these embodiments, the feature transformation selection module 230 includes a preliminary screening unit and an optimal selection unit;
the primary screening unit is used for carrying out characteristic primary screening on the high-dimensional characteristics to obtain primary screening characteristics;
and the optimal selection unit is used for optimally selecting the primary screening features by utilizing the feature use model to obtain an optimal feature combination.
In some of these embodiments, the primary screening unit is further configured to perform a grouping process on the high-dimensional features using chi-square binning, and determine a first grouping feature and a corresponding optimal grouping threshold;
converting the optimal grouping threshold value of each group of first grouping features by using evidence weight coding, and selecting a second grouping feature from the first grouping features based on a preset information value;
And performing multi-feature correlation screening analysis and stability evaluation on the second grouping feature to obtain a primary screening feature.
In some of these embodiments, the model building apparatus further comprises a determination module based on fig. 4;
and the determining module is used for determining an expression of the health and the prediction probability based on the influence weight of each feature output by the quantitative evaluation model on the operation health degree.
In some of these embodiments, the expression of health and predictive probability is:
wherein Score represents a health Score; p represents a prediction probability; a and B represent constants.
The above-described respective modules may be functional modules or program modules, and may be implemented by software or hardware. For modules implemented in hardware, the various modules described above may be located in the same processor; or the above modules may be located in different processors in any combination.
The embodiment also provides an enterprise operation health assessment method, and fig. 5 is a flowchart of the enterprise operation health assessment method according to an embodiment of the present application, as shown in fig. 5, where the flowchart includes the following steps:
Step S410, performing quantitative evaluation on the original basic data of the enterprise to be analyzed by using the quantitative evaluation model constructed by the model construction method of any one of the embodiments, so as to obtain a quantitative evaluation result of the enterprise to be analyzed.
Through the steps, the quantitative evaluation result can be output, and the accuracy of the evaluation result is improved.
The present embodiment is described and illustrated below by way of preferred embodiments.
Fig. 6 is a flowchart of the enterprise business health assessment method of the preferred embodiment. As shown in fig. 6, the enterprise operation health assessment method includes the following steps:
step S610, processing original basic data of an enterprise to obtain multidimensional features;
step S620, sample processing is carried out on the multidimensional features, and high-dimensional features of the renewal enterprise samples and the loss enterprise samples are selected;
step S630, performing feature transformation selection on the high-dimensional features to determine an optimal feature combination;
step S640, dividing the optimal feature combination into a training set and a testing set, taking a renewal enterprise and a loss enterprise as two kinds of target variables of training learning, and training and learning an original network model with supervised learning by combining preset sectional renewal rates to construct a quantitative evaluation model;
Step S650, carrying out quantitative evaluation on the original basic data of the enterprise to be analyzed by utilizing a quantitative evaluation model to obtain a quantitative evaluation result of the enterprise to be analyzed;
step S660, inputting the quantitative evaluation result into the expression of the health and the prediction probability to obtain the operation health score and the health degree of the enterprise to be analyzed.
Specifically, the method can be combined with a model training process to realize quantitative evaluation of the original basic data of the enterprise to be analyzed, so as to obtain a quantitative evaluation result of the enterprise to be analyzed; and inputting the quantitative evaluation result into an expression of the health and the prediction probability to obtain the business health score and the health degree of the enterprise to be analyzed.
Through the steps, the operation health score and the health degree of the enterprise to be analyzed can be directly obtained, so that the problems that the operation health condition of the enterprise is difficult to comprehensively quantify, the prediction accuracy is low and the stability is poor are solved, meanwhile, the problem that the enterprise's continuous fee will is difficult to predict in the SaaS service field is avoided, the dilemma that the operation health condition lacks interpretable reasons and cannot guide the landing of the fine operation action is solved, the health degree of the enterprise in the operation process is mastered early, and the integral continuous rate is improved. In this embodiment, the enterprise to be analyzed may be an enterprise or a merchant in the platform e-commerce field or other fields where the enterprise business health needs to be related, as long as the enterprise or merchant is in the field of using SaaS services.
The embodiment also provides an enterprise operation health assessment device, as shown in fig. 7, including: a quantization evaluation module 510;
the quantization evaluation module 510 is configured to perform quantization evaluation on the original basic data of the enterprise to be analyzed by using the quantization evaluation model constructed by the model construction method according to any one of the embodiments, so as to obtain a quantization evaluation result of the enterprise to be analyzed.
By the device, the quantitative evaluation result can be output, and the accuracy of the evaluation result is improved.
In some embodiments, the enterprise operation health assessment device further comprises a health calculation module based on fig. 7;
and the health calculation module is used for inputting the quantitative evaluation result into an expression of the health and the prediction probability to obtain the operation health score and the health degree of the enterprise to be analyzed.
There is also provided in this embodiment a computer device comprising a memory in which a computer program is stored and a processor arranged to run the computer program to perform the steps of any of the method embodiments described above.
Optionally, the computer device may further include a transmission device and an input/output device, where the transmission device is connected to the processor, and the input/output device is connected to the processor.
Alternatively, in the present embodiment, the above-described processor may be configured to execute the following steps by a computer program:
s1, processing original basic data of an enterprise to obtain multidimensional features;
s2, sample processing is carried out on the multidimensional features, and high-dimensional features of the renewal enterprise samples and the loss enterprise samples are selected;
s3, carrying out feature transformation selection on the high-dimensional features to determine an optimal feature combination;
and S4, dividing the optimal feature combination into a training set and a testing set, taking a renewal enterprise and a loss enterprise as two kinds of target variables of training and learning, and training and learning the original network model with supervised learning by combining with each preset sectional renewal rate to construct a quantitative evaluation model.
Alternatively, the processor may be arranged to perform the following steps by a computer program:
and carrying out quantitative evaluation on the original basic data of the enterprise to be analyzed by utilizing the quantitative evaluation model constructed by the model construction method of any embodiment, so as to obtain a quantitative evaluation result of the enterprise to be analyzed.
It should be noted that, specific examples in this embodiment may refer to examples described in the foregoing embodiments and alternative implementations, and are not described in detail in this embodiment.
Further, the model construction method provided in the above embodiment is combined; or, the enterprise operation health assessment method may also be implemented by providing a storage medium in this embodiment. The storage medium has a computer program stored thereon; the computer program, when executed by a processor, implements any of the model building methods of the above embodiments; or, an enterprise operation health degree assessment method.
It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to be limiting. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure in accordance with the embodiments provided herein.
It is to be understood that the drawings are merely illustrative of some embodiments of the present application and that it is possible for those skilled in the art to adapt the present application to other similar situations without the need for inventive work. In addition, it should be appreciated that while the development effort might be complex and lengthy, it will nevertheless be a routine undertaking of design, fabrication, or manufacture for those of ordinary skill having the benefit of this disclosure, and further having the benefit of this disclosure.
The term "embodiment" in this disclosure means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive. It will be clear or implicitly understood by those of ordinary skill in the art that the embodiments described in the present application can be combined with other embodiments without conflict.
The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the patent claims. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of the application should be assessed as that of the appended claims.
Claims (13)
1. A method of modeling, comprising:
processing the original basic data of the enterprise to obtain multi-dimensional characteristics;
Sample processing is carried out on the multidimensional features, and high-dimensional features of renewal enterprise samples and loss enterprise samples are selected;
performing feature transformation selection on the high-dimensional features to determine an optimal feature combination;
dividing the optimal feature combination into a training set and a testing set, taking the renewal enterprise and the loss enterprise as two classification target variables of training and learning, and training and learning the original network model with supervised learning by combining with each preset segmentation renewal rate to construct a quantitative evaluation model.
2. The method for constructing a model according to claim 1, wherein the performing sample processing on the multidimensional feature to select high-dimensional features of a renewal enterprise sample and a churn enterprise sample includes:
preprocessing the multi-dimensional characteristics according to a preset preprocessing rule;
and carrying out sample processing on the preprocessed multidimensional features, and selecting the high-dimensional features of the pay-out enterprise samples and the loss enterprise samples.
3. The method for constructing a model according to claim 2, wherein the sample processing the preprocessed multi-dimensional features, selecting high-dimensional features of the pay-out enterprise sample and the churn enterprise sample, includes:
Sample processing is carried out on the preprocessed multidimensional features according to a preset screening strategy, and high-dimension features of the output renewal enterprise samples and the loss enterprise samples are selected;
the screening strategy comprises a service life threshold and an enterprise type sample division principle.
4. A model building method according to any one of claims 1 to 3, wherein said feature transformation selection of the high dimensional features to determine an optimal feature combination comprises:
performing feature primary screening on the high-dimensional features to obtain primary screening features;
and optimally selecting the primary screening features by using a feature use model to obtain an optimal feature combination.
5. The method of modeling in accordance with claim 4, wherein said performing feature prescreening on the high-dimensional features to obtain prescreened features comprises:
grouping the high-dimensional features by using a chi-square box, and determining a first grouping feature and a corresponding optimal grouping threshold;
converting the optimal grouping threshold value of each group of the first grouping features by using evidence weight coding, and selecting a second grouping feature from the first grouping features based on a preset information value;
And performing multi-feature correlation screening analysis and stability evaluation on the second grouping feature to obtain a primary screening feature.
6. The model building method according to claim 4, characterized in that the method further comprises:
and determining an expression of the health and the prediction probability based on the influence weight of each feature output by the quantitative evaluation model on the operation health.
7. The model construction method according to claim 6, wherein the expression of the health and prediction probability is:
wherein Score represents a health Score; p represents a prediction probability; a and B represent constants.
8. An enterprise business health assessment method, comprising:
carrying out quantitative evaluation on the original basic data of an enterprise to be analyzed by using the quantitative evaluation model constructed by the model construction method according to any one of claims 1 to 7 to obtain a quantitative evaluation result of the enterprise to be analyzed.
9. The method of assessing the operational health of an enterprise of claim 8, further comprising:
and inputting the quantitative evaluation result into an expression of health and prediction probability to obtain the business health score and the health degree of the enterprise to be analyzed.
10. A model building apparatus, comprising: the device comprises a processing module, a sample processing module, a characteristic transformation selection module and a training learning module;
the processing module is used for processing the original basic data of the enterprise to obtain multidimensional features;
the sample processing module is used for processing the samples of the multidimensional features and selecting high-dimensional features of the renewal enterprise samples and the loss enterprise samples;
the feature transformation selection module is used for carrying out feature transformation selection on the high-dimensional features so as to determine an optimal feature combination;
the training learning module is used for dividing the optimal feature combination into a training set and a testing set, taking the renewal enterprise and the loss enterprise as two kinds of target variables of training learning, and combining preset sectional charge rates to perform training learning on the original network model with supervised learning so as to construct a quantitative evaluation model.
11. An enterprise business health assessment apparatus, comprising: a quantization evaluation module;
the quantitative evaluation module is configured to perform quantitative evaluation on the original basic data of the enterprise to be analyzed by using the quantitative evaluation model constructed by the model construction method according to any one of claims 1 to 7, so as to obtain a quantitative evaluation result of the enterprise to be analyzed.
12. A computer device comprising a memory and a processor, characterized in that the memory has stored therein a computer program, the processor being arranged to run the computer program to perform the steps of the model building method according to any of claims 1 to 7; or, performing the steps of the enterprise business health assessment method of any one of claims 8 to 9.
13. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the model building method according to any one of claims 1 to 7; or, the steps of implementing the enterprise business health assessment method of any one of claims 8 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210460085.6A CN117035472A (en) | 2022-04-28 | 2022-04-28 | Enterprise operation health assessment method, model construction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210460085.6A CN117035472A (en) | 2022-04-28 | 2022-04-28 | Enterprise operation health assessment method, model construction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117035472A true CN117035472A (en) | 2023-11-10 |
Family
ID=88600958
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210460085.6A Pending CN117035472A (en) | 2022-04-28 | 2022-04-28 | Enterprise operation health assessment method, model construction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117035472A (en) |
-
2022
- 2022-04-28 CN CN202210460085.6A patent/CN117035472A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108564286B (en) | Artificial intelligent financial wind-control credit assessment method and system based on big data credit investigation | |
| EP3803714A1 (en) | Systems and methods for decomposition of non-differentiable and differentiable models | |
| CN110956273A (en) | Credit scoring method and system integrating multiple machine learning models | |
| WO2020257782A1 (en) | Factory risk estimation using historical inspection data | |
| CN116187835A (en) | Data-driven-based method and system for estimating theoretical line loss interval of transformer area | |
| Hu et al. | Metric-free individual fairness with cooperative contextual bandits | |
| CN113011788B (en) | Emergency decision method for offshore traffic accident, terminal equipment and storage medium | |
| CN113674087A (en) | Enterprise credit rating method, apparatus, electronic device and medium | |
| US20210357699A1 (en) | Data quality assessment for data analytics | |
| CN113256325A (en) | Second-hand vehicle valuation method, system, computing device and storage medium | |
| CN113704389A (en) | Data evaluation method and device, computer equipment and storage medium | |
| WO2020257784A1 (en) | Inspection risk estimation using historical inspection data | |
| CN111738610A (en) | Public opinion data-based enterprise loss risk early warning system and method | |
| CN115730152A (en) | Big data processing method and big data processing system based on user portrait analysis | |
| CN117035472A (en) | Enterprise operation health assessment method, model construction method and device | |
| Deng et al. | A novel method for elimination of inconsistencies in ordinal classification with monotonicity constraints | |
| CN115115414A (en) | Second-hand car valuation method based on machine learning | |
| Thompson | Data mining methods and the rise of big data | |
| CN114092216A (en) | Enterprise credit rating method, apparatus, computer device and storage medium | |
| Mani et al. | An investigation of wine quality testing using machine learning techniques | |
| CN113064883A (en) | Method for constructing logistics wind control model, computer equipment and storage medium | |
| Tu et al. | A novel grey relational clustering model under sequential three-way decision framework | |
| Dixon et al. | A Bayesian approach to ranking private companies based on predictive indicators | |
| CN115423148B (en) | Agricultural machinery operation performance prediction method and device based on Ke Li jin method and decision tree | |
| Zanke et al. | Predictive Modelling for Insurance Pricing: A Comparative Analysis of Machine Learning Approaches |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |