CN115759460A - Method for predicting cooperation relationship between core enterprise and supplier based on convolutional neural network - Google Patents
Method for predicting cooperation relationship between core enterprise and supplier based on convolutional neural network Download PDFInfo
- Publication number
- CN115759460A CN115759460A CN202211528032.XA CN202211528032A CN115759460A CN 115759460 A CN115759460 A CN 115759460A CN 202211528032 A CN202211528032 A CN 202211528032A CN 115759460 A CN115759460 A CN 115759460A
- Authority
- CN
- China
- Prior art keywords
- data
- time sequence
- core enterprise
- neural network
- supplier
- 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
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the field of cooperative relationship prediction, and discloses a convolutional neural network-based method for predicting the cooperative relationship between a core enterprise and a supplier, which comprises the following steps: label definition, namely labeling historical data; a learning step, namely predicting the probability of termination of cooperation between suppliers and core enterprises in different industries in historical data within a certain time in the future, and completing learning; a data construction step, namely converting one-dimensional time sequence data into picture data which can be used as CNN input; a convolution processing step, inputting the picture data into a CNN model and carrying out convolution processing; a full-connection processing step, wherein the processed time sequence data and the non-time sequence data are input into a full-connection layer in the neural network together for processing; and a result output step of outputting a prediction result of the probability of the termination of the cooperation of the core enterprise and the supplier. The method and the device can predict the termination cooperation probability of the core enterprise and the supplier in a certain period of time in the future.
Description
Technical Field
The invention relates to the field of cooperative relationship prediction, in particular to a method for predicting the cooperative relationship between a core enterprise and a supplier based on a convolutional neural network.
Background
The supplier refers to an upstream enterprise providing raw materials for the core enterprise, one type of financial product in supply chain finance is supplier loan and is a financial service provided for the supplier, and the main admission condition and the credit granting logic of the credit mainly depend on the cooperation depth and the size of the supplier and the core enterprise. On one hand, the main business income of the supplier comes from the core enterprise, and when the core enterprise stops cooperating with the supplier in the future, the repayment capability of the core enterprise is directly reduced, so that bad accounts are caused; on the other hand, the core enterprise does not have interactive data with the supplier subsequently, management in loan is deficient, and the risk is difficult to continue to be controlled. Therefore, predicting whether a core enterprise and a supplier cease to collaborate at a future time is particularly important.
In order to solve the above problems, the prior art has several solutions: the method comprises the steps that firstly, pre-judgment is carried out by means of original running water data of core enterprises such as stock, sales and inventory and financial statements of provider enterprises and the like according to abundant field expert experience; the method has the defects of time consumption, large fluctuation of prediction accuracy, scarce people with abundant experience and certain subjectivity. Secondly, deriving key indexes based on original data interacted with the core enterprise, making a report according to a time cycle, and observing every day; the method has the disadvantages that effective indexes are difficult to create, invalid indexes can interfere with correct judgment, different people can read the indexes differently, and the subjectivity is still strong. Thirdly, after a large number of indexes are established according to original data, records of whether the cooperation between the core enterprise and the supplier is stopped at a certain time point in history are collected to be used as samples, simple statistical learning methods such as logistic regression are adopted to model the samples, appropriate variables are screened and input into a model, and coefficients are fitted to obtain an appropriate model; the method has the defects that a great amount of characteristic attempts are carried out by fine characteristic engineering, the characteristics used by a statistical model can only be subjected to simple slice statistical derivation, information in data is lacked, and the accuracy of the obtained prediction model is insufficient.
Disclosure of Invention
The invention aims to provide a method for predicting the cooperative relationship between a core enterprise and a supplier based on a convolutional neural network, so as to predict the termination cooperative probability of the core enterprise and the supplier in a certain period of time in the future.
In order to achieve the purpose, the invention adopts the following technical scheme: the method for predicting the cooperative relationship between the core enterprise and the supplier based on the convolutional neural network comprises the following steps:
a label definition step, wherein labels are carried out on historical data, and the labels comprise a termination cooperation label and an untermination cooperation label;
a learning step, namely predicting the probability of termination of cooperation between suppliers and core enterprises in different industries in historical data within a certain time in the future to obtain a historical prediction result and finish learning;
a sample construction step, namely constructing a supplier of a certain date and a corresponding core enterprise into a sample point;
a sample grouping step, namely segmenting the constructed sample points according to the industry to obtain sample groups, and respectively modeling the sample groups;
a data construction step, namely converting one-dimensional time sequence data into picture data which can be used as CNN input;
a convolution processing step, inputting the picture data into a CNN model and carrying out convolution processing;
a full-connection processing step, wherein the processed time sequence data and the non-time sequence data are input into a full-connection layer in the neural network together for processing;
and a result output step of outputting a prediction result of the probability of the termination of the cooperation of the core enterprise and the supplier.
The principle and the advantages of the scheme are as follows: in actual application, the learning step is used for predicting and learning the probability of termination of cooperation of suppliers and core enterprises in different industries in historical data within a certain time in the future to obtain historical prediction results, labeling the historical prediction results, obtaining multiple groups of historical prediction results in a long-time sequence, and completing model learning according to the historical data; the method comprises the steps that a date, a supplier and a core enterprise are constructed into a sample point in a one-to-one correspondence mode, and different industries have different specifications and standards, so that samples are grouped, the constructed sample point is segmented according to the industries, and different models are constructed according to the different industries, so that a prediction result is more professional and accurate; and a data construction step, namely converting one-dimensional time sequence data into picture data which can be used as CNN input, performing convolution processing, inputting the processed time sequence data and non-time sequence data into a full connection layer in a neural network together for processing, and finally obtaining a prediction result of the probability of the termination of cooperation of the core enterprise and the supplier.
The convolutional neural network can only process data in a two-dimensional picture format and can capture local important information of the picture data, and the technology of the convolutional neural network, which is commonly used for two-dimensional data such as images, is applied to the traditional table data by converting the one-dimensional table data into the two-dimensional picture data, so that the local important information capture of numerous original data between a supplier and a core enterprise is realized, the information in the original data is more comprehensively utilized, the probability prediction precision of the termination of cooperation is improved, and therefore, financial institutions can design different strategies in different scenes according to the probability of the termination of cooperation, the bad account risk is reduced, and credit loan is more accurately carried out; meanwhile, complicated characteristic engineering in the model building process is avoided, and the working efficiency is improved.
Preferably, as an improvement, the data constructing step includes:
a data standardization step, namely summarizing time sequence data into monthly data according to natural months, and cleaning non-time sequence data to obtain a standardized data format;
a time sequence data construction step, wherein one-dimensional time sequence data are sorted and arranged into data arranged in 36 rows by 1 columns according to time sequence; and converting the data arranged in 36 rows by 1 column into two-dimensional table data in 12 rows by 3 columns, wherein each column is 12 months, the data are sequentially arranged from top to bottom according to the time sequence, and the two-dimensional table data are normalized and converted into grayscale image data with the gray scale value of 0-1.
The technical effects are as follows: the original data are sorted and cleaned, one-dimensional table data are converted into two-dimensional picture data, conversion of the table data and the image data is achieved, and therefore the table data can be input into a convolutional neural network model to be processed. The processed two-dimensional data are arranged in 12 months in one year in sequence in each column, and are arranged in the same month in the sequential year in each row in sequence, so that the convolutional layers in the convolutional neural network can conveniently capture information such as the same ratio and the ring ratio.
Preferably, as an improvement, the data normalization step of aggregating the time-series data into monthly data by natural month includes aggregating data whose time unit is day and filling in missing data.
The technical effects are as follows: the time sequence data is processed into a standardized format, so that the influence on the prediction result caused by the error of the numerical value is avoided.
Preferably, as an improvement, in the data normalization step, one of a binning method, a clustering method and a regression method is used for cleaning the non-time-series data.
Preferably, as an improvement, the method further includes a network structure output step of outputting a network structure in which the input time-series data is subjected to convolutional layer processing but the non-time-series data is not added yet to be subjected to full link layer processing.
The technical effects are as follows: and the manager can analyze and judge the reliability of the model according to the network structure, so that the model is adjusted according to the reliability.
Preferably, as an improvement, the method further comprises a step of detecting the performance of the model, wherein the performance of the model is evaluated, and the evaluation index is an AUC index in the binary problem.
The technical effects are as follows: according to the AUC index value, the model performance can be judged, and the larger the AUC value is, the better the model performance is.
Preferably, as an improvement, the time-series data includes an amount of incoming goods, an amount of return goods, an amount of sales, an amount of stock, a reconciliation amount, and an amount of payment.
The technical effects are as follows: the receiving and paying conditions of the input amount, the return amount, the sales amount, the inventory amount, the account checking and the payment amount are researched according to the time sequence, and the probability of the future stop of the cooperation between the core enterprise and the supplier can be analyzed by capturing the key information of the indexes.
Preferably, as an improvement, the non-chronological data includes business-level registered capital, total assets, liabilities, and net profits.
The technical effects are as follows: the indexes of enterprise-level registered capital, total assets, liability and net profits are brought into the model together with the time sequence data for calculation, and the model can be established more accurately.
Drawings
FIG. 1 is a schematic flow chart of a convolutional neural network-based core enterprise and supplier partnership prediction method;
FIG. 2 is a two-dimensional tabular data example diagram of a convolutional neural network-based core enterprise and supplier partnership prediction method;
fig. 3 is a diagram illustrating an example of gray scale image data of a convolutional neural network-based core enterprise and supplier partnership prediction method.
Detailed Description
The following is further detailed by way of specific embodiments:
example one
The embodiment is basically as shown in the attached figure 1:
the method for predicting the cooperative relationship between the core enterprise and the supplier based on the convolutional neural network comprises the following steps:
and a label defining step of labeling the historical data, wherein the labels comprise a termination cooperation label and an un-termination cooperation label, and the termination cooperation label is 1 and the un-termination cooperation label is 0 by adopting the thought of two categories.
A learning step, namely predicting the probability of termination of cooperation between suppliers and core enterprises in different industries in historical data within a certain time in the future to obtain a historical prediction result and finish learning; the time length covered by the historical prediction result is set according to actual needs, for example, the probability of terminating the cooperation in the future 1 month is set and predicted, or the probability of terminating the cooperation in the future 2 months is set, even the longer the time window is, the greater the prediction difficulty is, and the probability of terminating the cooperation in the future 1 month is preferably predicted in the application; if the historical data of the past 10 years is learned, it is preferable that the time segment is set every three years in the present application, and if month 1 in 2003 is a prediction point, prediction is performed based on the historical data of month 1 in 2001 to month 12 in 2002, and similarly, learning is performed by pushing back by month.
A sample construction step, in which a supplier of a date and a corresponding core enterprise are constructed into a sample point, for example: a supplier tax reciept + a core enterprise tax reciept + a yyymm formatted date (e.g., 202108) as a record, i.e., a sample point.
A sample grouping step, namely segmenting the constructed sample points according to the industry to obtain sample groups, and respectively modeling the sample groups; because different core enterprises in different industries have larger difference, in order to make a prediction result more accurate and have pertinence, grouping modeling is needed, the sample is preferably segmented according to the national standard first-level industry standard in the application, and the first-level industry with enough samples is segmented according to the specific core enterprises.
A data construction step, namely converting one-dimensional time sequence data into picture data which can be used as CNN input; the data construction step comprises a data standardization step and a time sequence data construction step, and the time sequence data preferably comprises an incoming amount, a returned amount, a sales amount, an inventory amount, a reconciliation amount and a payment amount.
A data standardization step, namely summarizing time sequence data into monthly data according to natural months, wherein the data with a time unit of day is summarized, and missing data is filled; for example, the time unit of the sales is day, that is, there is sales data every day, and the sales data of each month needs to be summarized, that is, all the data of each month are summed to obtain monthly data; some data are not generated every month, for example, 1 purchase amount per month should be generated, and if one month of purchase amount data is missing in the time series for prediction, the missing data needs to be filled, and the missing data is filled to 0 in the present application; in the data standardization step, non-time sequence data are cleaned to obtain a standardized data format; the non-time series data preferably comprises registered capital, total assets, liabilities and net profits at the enterprise level, and the non-time series data preferably is cleaned by adopting one of a binning method, a clustering method and a regression method.
A time sequence data construction step, wherein one-dimensional time sequence data are sorted and arranged into data arranged in 36 rows by 1 columns according to time sequence; converting the data arranged in 36 rows by 1 column into two-dimensional table data in 12 rows by 3 columns, for example, 20210731 predicts whether the cooperation between the supplier a and the core enterprise P is ended in 2021 at 8 months, 20210731 is called observation day, the sample point is (supplier a, core enterprise P, 202108), the data to be acquired has time series data of 36 months before 20210731, namely, data of 201808 to 202107, such as monthly data of supplier a supplying to core enterprise P, monthly data of stock of the supplier a in the core enterprise P, monthly sales of commodities manufactured by the core enterprise P using the raw material of the supplier a, return data of the core enterprise P to the supplier a, and running data of time series such as reconciliation and payment between the core enterprise P and the supplier a, summarizing all monthly data according to a natural month, wherein the data at the moment is one-dimensional time sequence data which can be regarded as time sequence data of 36 rows × 1 column, the data is converted into two-dimensional data of 12 rows × 3 columns, as shown in fig. 2, each column is 12 months, the data are sequentially arranged from top to bottom according to time sequence, the leftmost column is 201808 to 201907, the middle column is 201908 to 202007, the rightmost column is 202008 to 202107, the data are horizontally arranged in the same month of different years according to the sequence, after the data are organized, the two-dimensional table data are normalized and converted into gray level image data of 0-1, and the row is like a gray level image at the moment, as shown in fig. 3: the larger the grayscale value, the darker the color, and the larger the corresponding numerical value. Similar conversions are made for the amount of incoming goods, the amount of return goods, the amount of sales, the amount of stock, the amount of reconciliation, the amount of payment, and the like. In CNN, each type of data is called a channel, and a plurality of channels are stacked together to form an RGB color image, and the time-series data is converted into an image and can be used as an input of CNN.
A convolution processing step, inputting the picture data into a CNN model and carrying out convolution processing; the CNN can perform image recognition and accurately capture the same-ratio and ring-ratio variations in time series data.
A full-connection processing step, wherein the processed time sequence data and the non-time sequence data are input into a full-connection layer in the neural network together for processing; in the application, the non-time sequence data is not specially processed, but only cleaned, so that a large amount of artificial characteristic engineering is avoided, the artificial characteristic engineering is time-consuming, information attenuation is caused in processing, and the prediction effect of the final model is influenced.
And a network structure output step, which is used for outputting the network structure after the convolution layer processing is carried out on the input time sequence data, but the non-time sequence data is not added for carrying out the full connection layer processing. By observing the number of parameters in the network structure, the complexity of the model can be known, the model can be adjusted and optimized in time, reference is provided for professionals in the same field, and the model can be trained by directly applying the same structure to try when similar problems are encountered.
And a result output step of outputting a prediction result of the probability of the termination of the cooperation of the core enterprise and the supplier. The output in this application is the probability that provider a and core enterprise P will terminate the cooperation within a month of the future.
The method also comprises a model performance detection step for evaluating the performance of the model, wherein the evaluation index is an AUC index in the two-class problem, the performance of the model can be judged according to the AUC index value, and the larger the AUC value is, the better the performance of the model is; in the present application, the area under the ROC (Receiver Operating Characteristic) curve, i.e. the AUC index value, reaches 0.936, while the highest AUC value in the prior art is 0.834.
Example two
The difference from the first embodiment is that, in the learning step, when the probability of termination of cooperation between the provider and the core enterprise within a certain time is predicted, the prediction duration is further determined by a decision importance coefficient, a variation coefficient and historical prediction accuracy, where the decision importance coefficient refers to an influence coefficient of the probability of termination of cooperation between the provider and the core enterprise due to a significant decision made by the provider and the core enterprise within the prediction time, the variation coefficient refers to a frequent or large business variation, financial variation, and personnel variation within the provider and the core enterprise, and the historical prediction accuracy is obtained by performing regression analysis on a historical prediction value and an actual result.
And respectively endowing the decision importance coefficient, the variation coefficient and the historical prediction accuracy with a basic value and a coefficient value, and during actual application, the learning module further determines the prediction duration according to the decision importance coefficient, the variation coefficient and the historical prediction accuracy in the learning time period. For example, the basic values of the decision importance coefficient, the variation coefficient and the historical prediction accuracy are 1, 1 and 1 respectively, the basic values are set according to actual needs, when the historical data from 1 month to 12 months in 2001 are predicted, the decision importance coefficient is 0.1, the variation coefficient is 0.2, the historical prediction accuracy is 0.7, the prediction time duration is 0.3 × 1-0.2 × 1+0.9 × 1=1, namely the prediction time duration is 1 month, and the prediction time duration values obtained by learning all the historical data are subjected to weighted average to obtain the final prediction time duration.
When a supplier and a core enterprise have important decisions and influence on the termination cooperation of the supplier and the core enterprise, the corresponding prediction time is longer, and when the supplier and the core enterprise have large changes, the corresponding prediction result changes, so that the prediction time is shortened, the historical prediction accuracy is higher, namely the prediction stability is higher, the corresponding prediction time is longer, and the prediction time is determined by introducing a decision importance coefficient, a change coefficient and the historical prediction accuracy, so that the prediction result is more accurate and stable.
The foregoing is merely an example of the present invention and common general knowledge in the art of designing and/or characterizing particular aspects and/or features is not described in any greater detail herein. It should be noted that, for those skilled in the art, without departing from the technical solution of the present invention, several variations and modifications can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicability of the patent. The scope of the claims of the present application shall be determined by the contents of the claims, and the description of the embodiments and the like in the specification shall be used to explain the contents of the claims.
Claims (8)
1. The method for predicting the cooperative relationship between the core enterprise and the supplier based on the convolutional neural network is characterized by comprising the following steps:
a label definition step, wherein labels are carried out on the historical data, and the labels comprise a termination cooperative label and an unterminated cooperative label;
a learning step, namely predicting the probability of termination of cooperation between suppliers and core enterprises in different industries in historical data within a certain time in the future to obtain a historical prediction result and finish learning;
a sample construction step, constructing a supplier of a certain date and a corresponding core enterprise into a sample point;
a sample grouping step, namely segmenting the constructed sample points according to the industry to obtain sample groups, and respectively modeling the sample groups;
a data construction step, namely converting one-dimensional time sequence data into picture data which can be used as CNN input;
a convolution processing step, namely inputting the image data into a CNN model and carrying out convolution processing;
a full-connection processing step, wherein the processed time sequence data and the non-time sequence data are input into a full-connection layer in the neural network together for processing;
and a result output step of outputting a prediction result of the probability of the termination of the cooperation of the core enterprise and the supplier.
2. The convolutional neural network-based core enterprise and provider partnership prediction method of claim 1, wherein the data construction step comprises:
a data standardization step, namely summarizing time sequence data into monthly data according to natural months, and cleaning non-time sequence data to obtain a standardized data format;
a time sequence data construction step, wherein one-dimensional time sequence data are sorted and arranged into data arranged in 36 rows by 1 columns according to time sequence; and converting the data arranged in 36 rows by 1 column into two-dimensional table data in 12 rows by 3 columns, wherein each column is 12 months, the data are sequentially arranged from top to bottom according to the time sequence, and the two-dimensional table data are normalized and converted into grayscale image data with the gray scale value of 0-1.
3. The convolutional neural network-based core enterprise and provider partnership prediction method of claim 2, wherein: in the data normalization step, the collecting of the time series data into monthly data according to a natural month includes collecting data of which the time unit is a day and filling missing data.
4. The convolutional neural network-based core enterprise and provider partnership prediction method of claim 2, wherein: in the data standardization step, the non-time sequence data are cleaned by one of a box separation method, a clustering method and a regression method.
5. The convolutional neural network-based core enterprise and provider partnership prediction method of claim 1, wherein: and a network structure output step, which is used for outputting the network structure after the convolution layer processing is carried out on the input time sequence data, but the non-time sequence data is not added for carrying out the full connection layer processing.
6. The convolutional neural network-based core enterprise and provider partnership prediction method of claim 1, wherein: the method also comprises a model performance detection step for evaluating the performance of the model, wherein the evaluation index is an AUC index in the two-class problem.
7. The convolutional neural network-based core enterprise and provider partnership prediction method of claim 1, wherein: the time series data includes the amount of incoming goods, the amount of return goods, the amount of sales, the amount of inventory, account checking and the amount of payment.
8. The convolutional neural network-based core enterprise and supplier partnership prediction method of claim 1, wherein: the non-chronological data includes registered capital, gross assets, liabilities, net profits at the enterprise level.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211528032.XA CN115759460A (en) | 2022-11-30 | 2022-11-30 | Method for predicting cooperation relationship between core enterprise and supplier based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211528032.XA CN115759460A (en) | 2022-11-30 | 2022-11-30 | Method for predicting cooperation relationship between core enterprise and supplier based on convolutional neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115759460A true CN115759460A (en) | 2023-03-07 |
Family
ID=85341969
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211528032.XA Pending CN115759460A (en) | 2022-11-30 | 2022-11-30 | Method for predicting cooperation relationship between core enterprise and supplier based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115759460A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116821636A (en) * | 2023-08-30 | 2023-09-29 | 湖南云滨信息技术有限公司 | Internet of things data acquisition, analysis and management system based on big data |
-
2022
- 2022-11-30 CN CN202211528032.XA patent/CN115759460A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116821636A (en) * | 2023-08-30 | 2023-09-29 | 湖南云滨信息技术有限公司 | Internet of things data acquisition, analysis and management system based on big data |
| CN116821636B (en) * | 2023-08-30 | 2023-11-14 | 湖南云滨信息技术有限公司 | Internet of things data acquisition, analysis and management system based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109829631B (en) | Enterprise risk early warning analysis method and system based on memory network | |
| CN114048436A (en) | Construction method and construction device for forecasting enterprise financial data model | |
| CN104778622A (en) | Method and system for predicting TPS transaction event threshold value | |
| CN112418738B (en) | Staff operation risk prediction method based on logistic regression | |
| CN111291925A (en) | Financial market prediction and decision-making system and method based on artificial intelligence | |
| CN117151345A (en) | Enterprise management intelligent decision platform based on AI technology | |
| CN112700324A (en) | User loan default prediction method based on combination of Catboost and restricted Boltzmann machine | |
| CN115759460A (en) | Method for predicting cooperation relationship between core enterprise and supplier based on convolutional neural network | |
| CN115145993A (en) | Railway freight big data visualization display platform based on self-learning rule operation | |
| Han et al. | Information technology investment and manufacturing worker productivity | |
| CN115205011A (en) | Bank user portrait model generation method based on LSF-FC algorithm | |
| Singha et al. | Application of different machine learning models for supply chain demand forecasting: comparative analysis | |
| Raheem et al. | Forecasting foreign exchange rate: Use of FbProphet | |
| KR20220067924A (en) | Loan regular auditing system using artificia intellicence | |
| CN117114812A (en) | Financial product recommendation method and device for enterprises | |
| CN116308494A (en) | Supply chain demand prediction method | |
| Yu et al. | Long-term and short-term effects of green strategy on corporate performance: evidence from Chinese listed companies | |
| CN116228403A (en) | Personal bad asset valuation method and system based on machine learning algorithm | |
| KR102499182B1 (en) | Loan regular auditing system using artificia intellicence | |
| CN115689331A (en) | Power transmission and transformation project quantity rationality analysis method based on MLP | |
| CN110264251B (en) | Data organization form for representing cash flow and prediction method based on multi-task learning | |
| CN112734577A (en) | Multicycle decision-making AI stock-frying robot platform | |
| CN114154617A (en) | Low-voltage resident user abnormal electricity utilization identification method and system based on VFL | |
| CN113850609A (en) | Customer management system, method, computer equipment and storage medium | |
| CN113837803A (en) | Second-hand mobile phone recycling price prediction algorithm based on multi-model fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |