Regional industry difference-based science and technology enterprise evaluation model construction method, equipment and medium

Abstract

The invention relates to a regional industry difference-based science and technology enterprise evaluation model construction method and device, and a medium. The method comprises the steps of determining a science and technology enterprise evaluation system; based on the index science and technology relevance and the data integrity information, dividing the science and technology enterprise evaluation indexes into a first type of indexes used for carrying out enterprise score prediction and a second type of indexes used for processing the first type of indexes which do not meet requirements; constructing an initial evaluation model based on a logistic regression model and expert experience; constructing a region plug-in sub-model based on a region heterogeneity index and an industry plug-in sub-model based on an industry heterogeneity index; and carrying out weighted fusion on the initial evaluation model, the region plug-in sub-model and the industry plug-in sub-model to obtain a final science and technology enterprise evaluation model. Compared with the prior art, the scientific and technical enterprises with industry differences in different regions can be accurately scored.

Description

Technical Field

[0001] The present invention relates to the field of bank loan evaluation, and in particular to a method, device and medium for constructing a technology-based enterprise evaluation model based on regional industry differences. Background Art

[0002] Due to the unique attributes of technology companies, namely "early investment, high growth, high returns, high risks and long cycles", financial institutions' value recognition and evaluation systems for technology companies are still not sound. The traditional loan logic of banks mainly relies on relevant data on the company's operating conditions, but in practice it has been found that this can no longer match the development of technology companies.

[0003] Due to regional and industry heterogeneity, there are large differences between different regions and industries. The existing unified evaluation model is difficult to apply to scenarios with different regional, industrial and customer characteristics. In addition, due to the rapid development of technology-based enterprises, experts are also developing their businesses by trial and error, resulting in insufficient authoritative expert experience. At the same time, the number of bad samples accumulated in recent years is not enough to support the establishment of an effective machine learning model. At the same time, due to the lack of evaluation models for technology-based enterprises, banks' credit marketing, risk control and other work cannot be carried out efficiently.

[0004] Therefore, there is an urgent need to design a more targeted and accurate method for constructing an evaluation model for technology-based enterprises. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a more targeted and accurate method, device and medium for constructing a technology-based enterprise evaluation model based on regional industry differences.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] According to a first aspect of the present invention, a method for constructing a technology-based enterprise evaluation model based on regional industry differences is provided, the method comprising:

[0008] Determine the evaluation system for technology-based enterprises;

[0009] Based on the scientific and technological relevance of indicators and data integrity information, the evaluation indicators of science and technology enterprises are divided into the first category of indicators used for enterprise score prediction and the second category of indicators used to deal with the first category of indicators that do not meet the requirements;

[0010] Construct an initial evaluation model based on logistic regression model and expert experience;

[0011] Construct a regional plug-in sub-model based on regional heterogeneity indicators, and an industry plug-in sub-model based on industry heterogeneity indicators;

[0012] The initial evaluation model is weightedly integrated with the regional plug-in sub-model and the industry plug-in sub-model to obtain the final technology-based enterprise evaluation model.

[0013] Preferably, the second category indicators used to process the first category indicators that do not meet the requirements are specifically: when the data missing degree of the first category indicators is greater than a preset situation, the corresponding second category indicators are used for fitting replacement, and the first category indicators after fitting replacement are used as input.

[0014] Preferably, the adopting of corresponding second-category indicators for fitting replacement specifically comprises: adopting corresponding second-category indicators for fitting replacement based on a multivariate linear regression algorithm.

[0015] Preferably, the construction of an initial evaluation model based on a logistic regression model and expert experience specifically includes:

[0016] Obtain the weight of each indicator selected from the first category of indicators based on expert experience as the first weight;

[0017] The logistic regression model is used to calculate the weight of each indicator selected from the first category of indicators as the second weight;

[0018] The first weight and the second weight are combined in a set ratio to obtain a third weight, and an initial evaluation model is constructed based on the third weight.

[0019] Preferably, the various indicators selected from the first category of indicators are specifically: the IV value of each indicator in the first category of indicators is calculated, and the required evaluation indicators are selected by combining expert experience.

[0020] Preferably, the construction of the regional plug-in sub-model based on the regional heterogeneity index and the industry plug-in sub-model based on the industry heterogeneity index further includes: updating the weights of the plug-in sub-models at set intervals or according to actual needs.

[0021] Preferably, the method further comprises preprocessing the technology-based enterprise evaluation index data, wherein the preprocessing comprises:

[0022] Missing value processing: Delete the data with missing rate greater than the set value, and use the EM missing data filling algorithm based on naive Bayes to fill in the missing values;

[0023] Outlier handling: Identify and replace outliers through the box plot method;

[0024] Data standardization: For business processing data, we collect the experience of business experts, divide the data into boxes and determine the box values; for other data, through data analysis, we decide to use min-max standardization or boxing for continuous data; for discrete variables, we use one-hot encoding or boxing.

[0025] Preferably, the method further comprises: grading the technology-based enterprise evaluation scores output by the technology-based enterprise evaluation model.

[0026] According to a second aspect of the present invention, there is provided an electronic device, comprising a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the program, any one of the methods described above is implemented.

[0027] According to a third aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, any one of the methods described above is implemented.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] 1) The present invention fully considers the characteristics of technology-based enterprises, establishes an initial evaluation model by combining machine learning modeling with expert experience, effectively solves the problem that technology-based enterprises are developing rapidly, expert experience is not authoritative enough, and the current samples are small and the modeling effect is poor. The regional plug-in sub-model and industry plug-in sub-model that consider regional and industry differentiation indicators are superimposed, and the regional industry sub-models can be componentized and modularized to achieve flexible combination of models, build a scoring model for technology enterprises that meets the needs of different regions and industries, and achieve comprehensive and accurate evaluation and scoring of technology-based enterprises.

[0030] 2) Based on the technological relevance of indicators and data integrity information, the present invention divides the evaluation indicators of technology-based enterprises into first-category indicators for enterprise score prediction and second-category indicators for processing first-category indicators that do not meet the requirements. When the data missing degree of the first-category indicators is greater than the preset situation, the corresponding second-category indicators are used for fitting replacement, and the first-category indicators after fitting replacement are used as input, which further makes up for the defects of indicators with strong technological relevance but large data missing amount, and improves the accuracy of subsequent scoring model prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a flow chart of the method of the present invention;

[0032] Figure 2 This is a schematic diagram of the evaluation system for technology-based enterprises;

[0033] Figure 3 This is a schematic diagram of the evaluation index data processing;

[0034] Figure 4 This is a schematic diagram of the application scenario of the technology enterprise evaluation model based on regional industry differences;

[0035] Figure 5This is a schematic diagram of the interface of the PC-based technology enterprise evaluation model application system; among them, (a) and (b) are schematic diagrams of multi-dimensional evaluation scores and overall evaluation levels, respectively. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0037] Example

[0038] The technology-based enterprises mentioned in this embodiment are mainly high-tech enterprises, technology-based SMEs, software certified enterprises, high-tech Zhongguancun enterprises, specialized and new small giant enterprises, specialized and new SMEs, national technology innovation demonstration enterprises, national manufacturing single champion enterprises, Science and Technology Innovation Board listed companies, Science and Technology Innovation Board to be listed companies, strategic emerging industries and other recognized customers, and also include enterprises with technological attributes.

[0039] like Figure 1 As shown, this embodiment provides a method for constructing a technology-based enterprise evaluation model based on regional industry differences, the method comprising:

[0040] S1. Determine the evaluation system for technology-based enterprises;

[0041] In this embodiment, a database of technology-based enterprises is established, and a three-level technology-based enterprise evaluation system framework is established from five dimensions: human capital, scientific research and innovation, social recognition, business performance, and industry status. A set of standardized models is formed, and N regional, industry, and customer group plug-ins are formed to form a "1+N" technology-based enterprise evaluation model matrix combination, such as Figure 2 shown.

[0042] S2. Based on the scientific and technological relevance of the indicators and the data integrity information, the evaluation indicators of science and technology enterprises are divided into the first category of indicators for enterprise score prediction and the second category of indicators for processing the first category indicators that do not meet the requirements; when the data missing degree of the first category indicators is greater than the preset situation, the corresponding second category indicators can be used for fitting replacement based on the multivariate linear regression algorithm, and the first category indicators after fitting replacement are used as input;

[0043] The first type of indicators are strong scientific content indicators, and the second type of indicators are weak scientific content indicators.

[0044] In this embodiment, when the first category of indicators is the number of patents in the strong scientific and technological indicators, the corresponding second category of indicators include the sample industries, the number of patent transfers, the industry-university-research cooperation, the number of other intellectual property rights, the number of trademarks, the basic pension insurance for urban employees, the amount of registered capital, the number of covered countries, and the number of software works in the weak scientific and technological indicators.

[0045] Based on the multivariate linear regression algorithm, the corresponding second-type indicators are used for fitting replacement, such as Figure 3 As shown, the following preprocessing process is also included:

[0046] 1) Missing value processing: Delete the data with missing rate greater than 50%, and fill in the missing values ​​based on the EM missing data filling algorithm of Naive Bayes.

[0047] 2) Outlier processing: Identify and replace outliers through the box plot method.

[0048] 3) Data standardization: For data commonly used in business processing, we collect the experience of business experts to divide the data into bins and determine the bin values. For other data, we analyze the data and decide whether to use min-max standardization or binning for continuous data, and one-hot encoding or binning for discrete variables.

[0049] For enterprises whose patent quantity is not 0 or empty, the original patent quantity is still used. For enterprises whose patent quantity is 0 or empty, the fitting result is used as the patent quantity. The actual patent quantity is combined with the fitted patent quantity to form complete patent quantity indicator data.

[0050] S3. Construct an initial evaluation model based on logistic regression model and expert experience, including:

[0051] Calculate the IV value of each indicator in the first category, and select the required evaluation indicators by combining expert experience;

[0052] Obtain the weight of each indicator selected from the first category of indicators based on expert experience as the first weight;

[0053] The logistic regression model is used to calculate the weight of each indicator selected from the first category of indicators as the second weight;

[0054] The first weight and the second weight are combined in a set ratio to obtain a third weight, and an initial evaluation model is constructed based on the third weight.

[0055] The data is verified based on the initialization model. As the positive and negative sample data gradually accumulate, more test set and validation set data are accumulated and introduced into the logistic regression model. The test set and validation set data are used to continuously perform testing, training and verification. Finally, based on the initial model, the original indicators are iteratively adjusted, and the importance of expert weights is gradually reduced.

[0056] S4, constructing a regional plug-in sub-model based on regional heterogeneity indicators and an industry plug-in sub-model based on industry heterogeneity indicators; wherein the weights of the plug-in sub-models are updated at set intervals or according to actual needs;

[0057] In view of the uneven development of science and technology companies in different regions and industries, the evaluation model fully considers the heterogeneity of regions and industries. On the basis of the initial evaluation model, a combined prediction model is constructed according to regional and industry characteristics, including regional plug-ins and industry plug-ins.

[0058] The regional plug-in is based on the Porter Diamond Model. It builds a regional market development potential index system that includes six aspects: regional endowment, regional output capacity, regional industrial structure, regional demand, regional innovation level, and regional governance policy, based on traditional economic statistics such as regional population density, regional import and export trade volume, and regional consumer goods retail sales, as well as multivariate big data such as power big data and night view satellite images. Based on this index system, combined with statistical indicators such as the loan quality of the region within the bank, the regional market development potential evaluation index is comprehensively calculated.

[0059] Table 1 List of regional evaluation plug-in indicators

[0060]

[0061]

[0062] Industry plug-in construction selects industry-specific indicators based on the characteristics of each industry.

[0063] S5. The initial evaluation model is weightedly integrated with the regional plug-in sub-model and the industry plug-in sub-model to obtain a combined model that takes industry heterogeneity into consideration, i.e., the final technology-based enterprise evaluation model. Specifically, the technology-based enterprise evaluation scores output by the technology-based enterprise evaluation model can be graded.

[0064] The present invention fully considers the characteristics of technology-based enterprises, establishes a basic evaluation model by combining machine learning modeling with expert experience, effectively solves the problem that technology-based enterprises are developing rapidly, expert experience is not authoritative enough, and there are currently few samples and poor modeling effects. It also superimposes special plug-ins that consider regional and industry characteristics, and componentizes and modularizes regional industry sub-models to achieve flexible combination of models, builds technology enterprise evaluation models that meet the needs of different regions and industries, and achieves comprehensive evaluation and scoring of technology-based enterprises. At the same time, the model results are applied to the entire process of pre-loan, mid-loan, and post-loan, enabling the entire loan chain, using RPA technology to automatically distribute marketing tasks, and using NLP technology to automatically generate credit reports, providing quantitative basis in multiple application scenarios such as marketing and customer development, loan quota determination, and risk control, and is scalable and replicable for other peers.

[0065] By establishing an evaluation model for technology-based enterprises and building a professional evaluation system for technology-based enterprises based on it, the model can be embedded in the bank's traditional marketing, credit, and risk control processes, improving business work efficiency, supporting the development of high-quality technology-based enterprises faster and better, innovating financial service models, promoting the two-way empowerment of technology and finance, and promoting high-level scientific and technological self-reliance. Through open banking and SaaS platforms, government agencies, peers and other institutions are provided with technology-based enterprise ratings and related information inquiries, breaking information asymmetry, making technology-based enterprise data visible, understandable, and usable, promoting all-round support for technology-based enterprises, and helping my country's science and technology industry develop with high quality. Evaluating and scoring technology companies across society and improving the portrait of technology-based companies are flexible and extensive, and can be expanded to multiple scenarios, such as Figure 4 As shown, specifically including:

[0066] (1) Marketing push

[0067] The evaluation scoring and rating query functions for technology-based enterprises are realized on mobile and PC terminals to form a portrait of technology-based enterprises. The growth, stability and continuous innovation capabilities of technology-based enterprises are judged based on the five dimensions of human capital, scientific research and innovation, operating performance, social recognition, and industry status, providing multi-dimensional insights and quantitative identification for the business expansion of technology-based enterprises.

[0068] (2) Calculation of estimated amount

[0069] The evaluation model automatically generates a marketing intention credit line for technology-based enterprises based on information such as their score and rating, growth stage, operating income, financing amount, and credit line of approved customers. This helps branches expand their marketing and is used to attract target customers to the bank to conduct credit business. The actual credit line is subject to the credit approval line.

[0070] (3) Adjustment of quota

[0071] The evaluation model results support the development of “small, light and fast” products, and are embedded in customer access, loan amount determination, credit enhancement and other links through interface calls and other forms.

[0072] (4) Evaluation report generation

[0073] Generate a technology-based enterprise evaluation report with one click. The evaluation results of a single household of a technology-based enterprise model and the specific sub-item data content can be exported in the form of a report with one click through the "Generate a technology-based enterprise evaluation report with one click" function. The report can be used to support option selection, pre-examination review, credit application, etc., to increase efficiency and reduce burden for front-line account managers, provide reference for credit approval, and improve overall business efficiency. Figure 5 shown.

[0074] Support the writing of credit reports. Refer to the evaluation model results and evaluation reports to assist in generating credit reports in the form of automatic generation + online supplementary recording, and explore the optimization of credit processes and credit report templates for technology-based enterprises.

[0075] (5) External Services

[0076] QR code for customers: whitelist high-quality customers through evaluation model results. The target customers can be provided with QR codes by the account manager. By scanning, the system can query their own evaluation results and match the marketing intention quota.

[0077] Customer channels: For our own service channels, we can form customer portraits through the underlying model calculations, support query evaluation results and marketing intention quotas, and adapt financial products that meet the needs. For customers introduced through online channels, we will accurately adapt banking products to customers based on the model evaluation results and product recommendation rules, support customers to apply for products online, or be guided by account managers to use offline products.

[0078] Third-party platform: By cooperating with a third-party platform, the relevant content of the professional evaluation system for technology-based enterprises can be embedded in the business decision-making processes of third parties such as financial institutions, governments, and investment institutions, support the supplementary introduction of relevant data, and combine the "1+N" technology-based enterprise evaluation model to regain the results, form enterprise portraits and insight reports, support online query and other services, and support technology-based enterprise identification, screening, due diligence, review and other applications.

[0079] (6) Formation index

[0080] The evaluation results of the technology-based enterprise evaluation model can be combined with the enterprise's scoring situation to simultaneously form an industry, region, and customer group technological innovation capability evaluation index to support region, industry, and customer group tracking, trend analysis, and distribution analysis.

[0081] (7) Risk prevention and control

[0082] Based on monitoring changes in enterprise score ratings, rule strategies are designed as early warning references to extract and form risk warning signals to support risk prevention and control.

[0083] The electronic device of the present invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) or loaded from a storage unit into a random access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The CPU, ROM and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0084] Multiple components in the device are connected to the I / O interface, including: input units, such as keyboards, mice, etc.; output units, such as various types of displays, speakers, etc.; storage units, such as disks, optical disks, etc.; and communication units, such as network cards, modems, wireless communication transceivers, etc. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunication networks.

[0085] The processing unit performs the various methods and processes described above, such as methods S1 to S5. For example, in some embodiments, methods S1 to S5 may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via a ROM and / or a communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of methods S1 to S5 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S5 in any other appropriate manner (e.g., by means of firmware).

[0086] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chip systems (SOCs), load programmable logic devices (CPLDs), and the like.

[0087] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0088] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0089] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for constructing a technology-based enterprise evaluation model based on regional industry differences, characterized in that: The method includes: Determine the evaluation system for technology-based enterprises; Based on the scientific and technological relevance of indicators and data integrity information, the evaluation indicators of science and technology enterprises are divided into the first category of indicators used for enterprise score prediction and the second category of indicators used to deal with the first category of indicators that do not meet the requirements; Construct an initial evaluation model based on logistic regression model and expert experience; Construct a regional plug-in sub-model based on regional heterogeneity indicators, and an industry plug-in sub-model based on industry heterogeneity indicators; The initial evaluation model is weightedly integrated with the regional plug-in sub-model and the industry plug-in sub-model to obtain the final technology-based enterprise evaluation model.

2. The method for constructing a technology-based enterprise evaluation model based on regional industry differences according to claim 1 is characterized in that: The second type of indicators used to process the first type of indicators that do not meet the requirements are specifically: when the data missing degree of the first type of indicators is greater than a preset situation, the corresponding second type of indicators are used for fitting replacement, and the first type of indicators after fitting replacement are used as input.

3. The method for constructing a technology-based enterprise evaluation model based on regional industry differences according to claim 2 is characterized in that: The adopting of the corresponding second-category index for fitting replacement specifically comprises: adopting the corresponding second-category index for fitting replacement based on a multivariate linear regression algorithm.

4. The method for constructing a technology-based enterprise evaluation model based on regional industry differences according to claim 1 is characterized in that: The construction of the initial evaluation model based on the logistic regression model and expert experience specifically includes: Obtain the weight of each indicator selected from the first category of indicators based on expert experience as the first weight; The logistic regression model is used to calculate the weight of each indicator selected from the first category of indicators as the second weight; The first weight and the second weight are combined in a set ratio to obtain a third weight, and an initial evaluation model is constructed based on the third weight.

5. The method for constructing a technology-based enterprise evaluation model based on regional industry differences according to claim 4 is characterized in that: The various indicators selected from the first category of indicators are specifically: calculating the IV value of each indicator in the first category of indicators, and selecting the required evaluation indicators by combining expert experience.

6. The method for constructing a technology-based enterprise evaluation model based on regional industry differences according to claim 1 is characterized in that: The construction of the regional plug-in sub-model based on the regional heterogeneity index and the industry plug-in sub-model based on the industry heterogeneity index also includes: updating the weights of the plug-in sub-models at set intervals or according to actual needs.

7. The method for constructing a technology-based enterprise evaluation model based on regional industry differences according to claim 1 is characterized in that: The method further includes preprocessing the technology-based enterprise evaluation index data, and the preprocessing includes: Missing value processing: Delete the data with missing rate greater than the set value, and use the EM missing data filling algorithm based on naive Bayes to fill in the missing values; Outlier handling: Identify and replace outliers through the box plot method; Data standardization: For business processing data, we collect the experience of business experts, divide the data into boxes and determine the box values; for other data, through data analysis, we decide to use min-max standardization or boxing for continuous data; for discrete variables, we use one-hot encoding or boxing.

8. The method for constructing a technology-based enterprise evaluation model based on regional industry differences according to claim 1 is characterized in that: The method also includes: grading the technology-based enterprise evaluation scores output by the technology-based enterprise evaluation model.

9. An electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the program, the method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images