WO2022075915A1

WO2022075915A1 - Method and system for credit assessment

Info

Publication number: WO2022075915A1
Application number: PCT/SG2020/050562
Authority: WO
Inventors: Rajat Goswami PATIT PABAN GOSWAMI
Original assignee: Hitachi, Ltd.
Priority date: 2020-10-06
Filing date: 2020-10-06
Publication date: 2022-04-14

Abstract

An aspect of the disclosure relates to a method for determining a credit score associate with an entity, the method including: acquiring first financial data associate with an entity to be assessed; acquiring benchmark financial data associated with benchmark entities An aspect of the disclosure relates to a system configured to carry out the method, including: an input interface configured to receive the first financial data and the benchmark financial data; a computer memory configured to store the clusters; a processor configured to carry out: the extracting and the segregating of relevant data from the benchmark financial data into clusters; the analyzing and generating a benchmark of at least one of the clusters; the determining a first credit score of the entity based on the first financial data; and the calculating an aggregated credit score based on the first credit score and the benchmark.

Description

METHOD AND SYSTEM FOR CREDIT ASSESSMENT

TECHNICAL FIELD

[0001] An aspect of the disclosure relates to a method for determining a credit score associate with an entity. Another aspect of the disclosure relates to a system configured to carry out the method a method for determining a credit score associate with an entity. Another aspect of the disclosure relates to a non-transitory computer-readable medium storing computer executable code.

BACKGROUND

[0002] Current credit scoring systems’ efficiency is low for unbanked or underbanked merchants as available credit history details are be limited. As a result, many small and medium enterprises (SMEs) may not be able to access to credit from lenders. Lenders may be unable to obtain a sufficiently complete financial picture of a smaller merchant, and time spent on the credit assessment of an SME may be unprofitable for the lender since loan sizes are relatively small. Credit score systems that utilize artificial intelligence (A. I.) or machine learning models on the limited available credit history details may suffer from yielding poor results and results with an intrinsic increased credit risk. Thus, neither classical, nor A. I. systems are suitable for credit scoring of small and medium enterprises, and improved credit scoring system and method are needed.

SUMMARY

[0003] An aspect of the disclosure relates to a method for determining a credit score associate with an entity. The method may include acquiring first financial data associate with an entity to be assessed. The method may include acquiring benchmark financial data associated with benchmark entities. The method may include extracting and segregating relevant data from the benchmark financial data into clusters. The method may include analyzing and generating a benchmark of at least one of the clusters. The method may include determining a first credit score of the entity based on the first financial data; and may further include calculating an aggregated credit score based on the first credit score and the benchmark. [0004] An aspect of the disclosure relates to a system configured to carry out the method. The system may include an input interface configured to receive the first financial data and the benchmark financial data. The benchmark financial data may be anonymized, i.e., anonymized benchmark financial data. The system may include a computer memory configured to store the clusters. The system may include a processor configured to carry out the method in accordance with various embodiments.

[0005] An aspect of the disclosure relates to a non-transitory computer-readable medium storing computer executable code including instructions for determining a credit score according to the method of in accordance with various embodiments.

[0006] An aspect of the disclosure relates to a computer executable code including instructions for determining a credit score according to the method in accordance with various embodiments.

BRIEF DESCRIPTION OF THE DRAWINGS

[0007] The invention will be better understood with reference to the detailed description when considered in conjunction with the non-limiting examples and the accompanying drawings, in which:

- FIG. 1 shows a flowchart of a method for determining a credit score in accordance with various embodiments;

- FIG. 2 shows a flowchart of data being received by the Input and/or Output (I/O) interface, in accordance with various embodiments;

- FIG. 3 shows a flowchart of the data extraction and segregation, in accordance with various embodiments;

- FIG. 4 shows a flowchart of the benchmark analysis, in accordance with various embodiments;

- FIG. 5 shows a flowchart of the credit scoring determination, in accordance with various embodiments;

- FIG. 6 shows a table listing exemplary input data used for the determination of the aggregated credit score in accordance with various embodiments;

- FIG. 7 shows a table of an exemplary output of the determination of the aggregated credit score in accordance with various embodiments; - FIG. 8 shows a flowchart for the decision process, in accordance with various embodiments;

- FIG. 9 shows a schematic system in accordance with various embodiments;

- FIG. 10 shows an exemplary interface for input of financial data associate with an entity to be assessed by a user, in accordance with various embodiments;

- FIG. 11 shows on the right side, an exemplary summary data including an extracted information extracted from the consolidated balance sheets, in accordance with various embodiments;

- FIG. 12 shows an exemplary summary report with the credit decision in accordance with various embodiments; and

- FIG. 13 shows an exemplary flowchart of a method for determining a credit score in accordance with various embodiments.

DETAILED DESCRIPTION

[0008] The following detailed description refers to the accompanying drawings that show, by way of illustration, specific details and embodiments in which the disclosure may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the disclosure. Other embodiments may be utilized and structural, and logical changes may be made without departing from the scope of the disclosure. The various embodiments are not necessarily mutually exclusive, as some embodiments can be combined with one or more other embodiments to form new embodiments.

[0009] Embodiments described in the context of one of the systems or methods are analogously valid for the other systems or methods. Similarly, embodiments described in the context of a system are analogously valid for a method, and vice-versa.

[0010] Features that are described in the context of an embodiment may correspondingly be applicable to the same or similar features in the other embodiments. Features that are described in the context of an embodiment may correspondingly be applicable to the other embodiments, even if not explicitly described in these other embodiments. Furthermore, additions and/or combinations and/or alternatives as described for a feature in the context of an embodiment may correspondingly be applicable to the same or similar feature in the other embodiments.

[0011] In the context of various embodiments, the articles “a”, “an” and “the” as used with regard to a feature or element include a reference to one or more of the features or elements. [0012] As used herein, the term “and/or” includes any and all combinations of one or more of the associated listed items.

[0013] According to various embodiments, a method for determining a credit score associate with an entity including acquiring financial data associate with an entity to be assessed is disclosed. The credit score may be or be based on the aggregated credit score as explained herein. The method may be carried out by at least one processor. The entity to be assessed may also be named a credit requestor or a borrower. The method may include acquiring first financial data associate with an entity to be assessed. First financial data may include the financial statements of the borrower. The method may include acquiring benchmark financial data associated with benchmark entities, said benchmark financial data may be anonymized. Benchmark financial data may include financial statements of the benchmark entities, for example, peers of the borrower, and/or market leaders having comprehensive financial statements. The method may include extracting and segregating relevant data from the benchmark financial data into clusters. The method may include analyzing and generating a benchmark of at least one of the clusters. The method may include determining a first credit score of the entity based on the first financial data; and may further include calculating an aggregated credit score based on the first credit score and the benchmark. In some embodiments, acquiring benchmark financial data associated with benchmark entities may mean acquiring benchmark financial data which has already been anonymized.

[0014] In accordance with various embodiments, anonymization of benchmark financial data and determining the credit score may be carried out at different locations, for example the benchmark financial data may be anonymized at an accounting firm and the aggregated credit score may be determined by a money lender.

[0015] According to various embodiments, a method for determining a credit score is a computer implemented method. According to various embodiments, a database is implemented which is configured to store the financial statements obtained, this database may act as a repository to store anonymous financial statements of one or more entities, for example the benchmark financial statements, and financial statements of the entity to be assessed, for example, included in the first financial data. The financial statements may include one or more of balance sheet(s), cash flow statement(s), income statement(s), and others. A balance sheet may include line items related to assets and liabilities details. A cash flow statement may include line items related to cash flow from operations, investing, and financing activities details. An income statement may include line items related to revenue and expenses.

[0016] According to some embodiments extracting relevant data from the financial data may include natural language processing (NLP by a trained NLP processor. Alternatively or in addition, extracting relevant data from the financial data may include dark data analysis. In one example, anonymous data may be obtained from auditing and/or accounts company which have performed the anonymization, thus anonymization may not be necessary in a subsequent method step. Anonymous data may include anonymized dark data. Dark data may be defined as the information assets organizations collect, process and store during regular business activities, but may fail to use for other purposes (for example, analytics, business relationships and direct monetizing). Some embodiments according to the present disclosure may make use of dark data from companies like accounting firm, admin firms, or secretariat firms. Thus, if benchmark financial data is acquired from dark data, a dark data analysis step may be included in the method for determining a credit score.

[0017] According to some embodiments, the method may include a step of data anonymization. For example, data may be anonymized before it is received by the or, when benchmark financial data is received without being anonymized, extracting relevant data from the benchmark financial data may include a step of data anonymization. In the anonymization step, private data from the financial data may be not carried over to the relevant data. Private data may be identified according to pre-determined private data identification criteria. This anonymization may be implemented with natural language processing (NLP) for identifying private data and/or distinguishing private data from the financial data required for the credit analysis.

[0018] According to various embodiments, segregating the relevant data into clusters may be carried out by a trained classifier. The classifier may be trained using a training dataset including training financial statements and the respective expected classification result into a cluster, e.g. the respective industry sector. The industry sector and, consequently the classes, may include one or more of: retail, food and beverages, transportation, agriculture, chemical industry, construction, education, financial services, a combination thereof. The training may continue until a convergence criterion between the class output and the expected class output is met. The accuracy of the trained classifier may be determined by classifying test data (which is similar to the training data but has not been used for training) with the trained classifier. Training may be supervised, alternatively, training may be unsupervised, or training may be supervised and unsupervised. According to various embodiments, the trained classifier may be configured to classify the relevant data into cluster according to a set of classes, the set of classes including industry sector classes including one or more of: retail, food and beverages, transportation, agriculture, chemical industry, construction, education, financial services, a combination thereof.

[0019] According to various embodiments, an NLP processor is an electronic processor which may be implemented on a computer. Further, a classifier (e.g., a trained classifier) is an electronic classifier which may be implemented on a computer.

[0020] According to various embodiments, analyzing at least one of the clusters and generating a benchmark of at least one of the clusters is carried out for time frames of different duration.

[0021] According to some embodiments, determining a first credit score may include generating an A. I. score by a trained neural network, the A. I. score may be a prediction for example, a future score prediction. According to some embodiments, determining a first credit score may include generating a traditional score by a deterministic approach, such as using a score determination circuit. According to some embodiments, determining a first credit score may include generating an A. I. score by a trained neural network and may further include generating a traditional score by a deterministic approach, such as using a score determination circuit. The first credit score may include one or both of the A. I. score and the traditional score, and one or both of the A. I. score and the traditional score may be used in determining the aggregated credit score.

[0022] According to various embodiments, analyzing and generating a benchmark of each of the clusters may include compiling a benchmark summary data for each of the clusters and generating the benchmark from the benchmark summary data.

[0023] According to various embodiments, calculating an aggregated credit score may include aggregating the benchmark with the first credit score, wherein the first credit score may be weighted by a first pre-determined coefficient and the benchmark may be weighted by a second pre-determined coefficient. For example, the aggregated credit score may be calculated as Al X benchmark score + B X first credit score , wherein Al and B are predetermined weights. In another example, wherein the first credit score includes both of the A. I. score and the traditional score, the aggregated credit score may be calculated as Al X benchmark score + Bl x traditional score + Cl x A. I. score, wherein Al, Bl, and Cl are pre-determined weights, for example, Al = 0.3, Bl = 0.5, Cl = 0.2. The benchmark score may be calculate as avg(inventory turnover) + avg(ROCE)+ avg(current ratio), ‘avg’ meaning average, and ‘ROCE’ meaning Return On Capital Employed, but the disclosure is not limited thereto. In accordance with various embodiments, all scores may be normalized and the sum of the weights may be normalized, e.g., normalization may be to 1 so that the aggregated credit score is in the range from 0 to 1.

[0024] According to various embodiments, a system is envisaged configured to carry out the method in accordance with various embodiments. The system may include an input interface configured to receive the first financial data and the benchmark financial data. The benchmark financial data may be anonymized. The system may include a computer memory configured to store the clusters. The system may include a processor configured to carry out one or more of the: the extracting and the segregating of relevant data from the benchmark financial data into clusters; the analyzing and generating a benchmark of at least one of the clusters; the determining a first credit score of the entity based on the first financial data; the calculating an aggregated credit score based on the first credit score and the benchmark.

[0025] According to various embodiments, the system may further include a user input interface, and may further include a user output interface, such as a presentation unit.

[0026] According to various embodiments, a non-transitory computer-readable medium is envisaged storing computer executable code including instructions for determining a credit score according to the method in accordance with various embodiments.

[0027] According to various embodiments, a computer executable code is envisaged including instructions for determining a credit score according to the method in accordance with various embodiments.

[0028] FIG. 1 shows a flowchart of a method 100 for determining a credit score in accordance with various embodiments. The method may include obtaining financial data from the entity to be assessed and generating corresponding a first credit score. The method may include obtaining benchmark financial data from benchmark entities and generating corresponding benchmark credit score. The method may include generating an aggregated credit score based on the first credit score and the benchmark credit score. In some embodiments the benchmark financial data may be anonymized and obtained as anonymized benchmark financial data. [0029] According to various embodiments, the method 100 for determining a credit score associate with an entity may include acquiring first financial data 142 associate with an entity to be assessed. For example, the entity may be a Small or medium-sized enterprise (SME). The method may further include determining a first credit score 144 of the entity based on the first financial data, for example, the A. I. score and/or the traditional score.

[0030] According to various embodiments, the method may include acquiring the benchmark financial data 110 associated with benchmark entities. The method may further include extracting and segregating 120 relevant data from the benchmark financial data into clusters. The method may further include analyzing 130 and generating a benchmark of at least one of the clusters. The aggregated credit score may be calculated 150 based on the first credit score and the benchmark credit score. The method may further include a decision logic, for deciding 160 whether the credit is approved or rejected based on the aggregated credit score. Details of the method steps will be illustrated in the following.

[0031] FIG. 2 shows a flowchart of data being received by the Input and/or Output (VO) interface, in accordance with various embodiments. According to various embodiments, acquiring benchmark financial data 110 associated with benchmark entities may include receiving financial data via an VO interface 111 as shown in the flowchart of FIG. 2 by way of example. For example, the VO interface may be a network interface. The financial data may be stored in a database 112, which may be stored in computer memory. The financial data may include data related to one or more of balance sheet(s) 114, cash flow report(s) 116, income statement(s) 118, and others. According to various embodiments, the database 112 may be configured to and may act as a repository of financial statements, which may include one or more of balance sheet(s) 114, cash flow(s) 116, income statement(s) 118. In some embodiments, the benchmark financial data may already be anonymized.

[0032] As different companies follow different terms in the financial data (e.g., financial statements), a knowledge based constructor (KBC) may be used to populate the data base with the relevant financial information. One example of a KBC which may be used in accordance with various embodiments is Fonduer, which is a Python package and framework for building knowledge base construction applications from richly formatted data. The Fonduer package was found to provide a high accuracy and be able to extract different variations of line items from financial data. [0033] According to various embodiments, the extracting and segregating 120 relevant data from the benchmark financial data into clusters may include retrieving the financial data stored in the database 112, and processing the retrieved data with a segregator engine, e.g., step 120 as shown in the flowchart of FIG. 3 by way of example. According to various embodiments, extracting and segregating the relevant data into clusters may include, extracting information 122, clustering by industry 124 and storing the information in a database 126. Segregating the relevant data into clusters 124 may be carried out by a segregator engine, which may include a trained classifier. The segregator engine (e.g., in step 124), extracts information and stores the information in database 126 by industry.

[0034] The extracted information may be classified using supervised machine learning models, which are trained on actual industry financial statements. The anonymous financial statements after classification could be of industry specific small medium enterprises which may be categorized into, e.g., retail, food and beverages, transportation, agriculture, chemical industry, construction, education, financial services, a combination thereof, and others. According to some embodiments, the machine learning could be an unsupervised machine learning approach. For example, a k-means clustering algorithm may be utilized and data showing similar trends of variance may be grouped together into clusters. A combination of both supervised and unsupervised may also be utilized for better accuracy. The extracted data may be stored in database 126.

[0035] According to some embodiments extracting relevant data from the benchmark financial data may include a step of data anonymization wherein private data from the financial data may be not carried over to the relevant data, wherein private data may be identified according to pre-determined private data identification criteria. This may be used when benchmark financial data is not received in non-anonymized form.

[0036] According to various embodiments, the analyzing 130 and generating a benchmark of at least one of the clusters may include analyzing 130 each of the clusters, and may further include compiling a summary data for each of the clusters, the analyzing 130 and generating a benchmark of at least one of the clusters may include processing the information from database 126 by a benchmark analyzer, as shown in the flowchart of FIG. 4, step 130, by way of example. The data benchmark analyzer may create a benchmark for each cluster, which benchmark is stored in 132. According to various embodiments analyzing 130 least one of the clusters and generating a benchmark score of at least one of the clusters is carried out for time frames of different duration, for example 6 months, 12 months, or 24 months. An exemplary benchmark score, as illustrated in FIG. 4 is the inventory turnover of 20, 10, and 16 days for 6 months, 12 months, and 24 months respectively.

[0037] The benchmark analyzer is configured to analyze database 126 to create industry specific benchmark scores. The benchmark scores may be based on, or may be, financial ratios. The benchmark scores may be calculated using average, or median, or others, of the financial ratios across industry segments or clusters. Examples of financial ratios are the Current Ratio which is defined as the ratio of Current Assets to Current Liabilities, however the disclosure is not limited thereto. Customizable data fields may be stored in the database and the benchmark analyser may be configured to use these customizable data fields to calculate the financial ratios. The benchmark scores may be calculated using average, or median, or others, of the ROCE across industry segments or clusters. The benchmark scores may be calculated using average, or median, or others, of the inventory turnover across industry segments or clusters. The benchmark analyzer may be configured with formulas written as logic to calculate the financial ratios, ROCE, or inventory turnover. The disclosure is not limited to calculating only financial ratios, ROCE, or inventory turnover, and other values could be used. According to some embodiments, data bases 126 and/or 132 may be configured to store customizable data fields to allow for calculation of values other than financial ratios, ROCE, or inventory turnover. Since the aggregated credit score is based (in part) on the benchmark score, the benchmark scores may be used to determine the credit worthiness of new merchant with limited financial information.

[0038] FIG. 5 shows a flowchart of the credit scoring determination, in accordance with various embodiments. According to various embodiments, determining a first credit score 144 may include generating an A. I. score 145 by a trained neural network. Alternatively or in addition, determining a first credit score 144 may include generating a traditional score 146, for example, by a deterministic score determination circuit.

[0039] FIG. 6 shows a table listing exemplary input data used for the determination of the aggregated credit score in accordance with various embodiments. The left column shows the data name and the right column indicates the method step where the data originates. Shown are the A. I score produced in determining credit score 144, the traditional score produced in determining credit score 144, and the benchmark score, which may be produced by the benchmark analyzer 130. Weights, such as Al, Bl, and Cl may be pre-determined, for example, input by a user. Variations of the shown data are also envisaged, for example, less or more data may be used for the determination of the aggregated credit score. In some embodiments the benchmark score may be provided by a third party (e.g. a bank or a consulting firm).

[0040] FIG. 7 shows a table of an exemplary output of the determination of the aggregated credit score in accordance with various embodiments. According to various embodiments calculating an aggregated credit score 150 may include aggregating the benchmark with the first credit score, wherein the first credit score may be weighted by a first pre-determined coefficient and the benchmark may be weighted by a second pre-determined coefficient. For example, using the input data as shown in FIG. 6, the aggregated credit score may be determined in aggregated scoring 150 as Al X benchmark score + Bl X traditional score + Cl x A. I. score.

[0041] FIG. 8 shows a flowchart of the credit decision process 150, in accordance with various embodiments. According to various embodiments, the decision step 150 may include a threshold checking step 152 and a decision step 154, as shown in the flowchart of FIG. 8 by way of example. The threshold checking step 152 may be configured to compare a predetermined threshold with the aggregated score, the information whether the comparison meets the criterium (e.g., aggregated score is equal or over threshold or aggregated score does not meet threshold) may be presented to a user, for example in a credit decision sheet, a graphic user interface, etc. In some embodiments, the method may accept a further user input to approve or reject the credit request. Alternatively or in addition, the method may automatically decide or suggest a result of the credit request, for example, the credit request may be accepted in cases were the aggregated score is equal or over the pre-determined threshold and the credit score may be rejected in cases were the aggregated score is smaller than the pre-determined threshold. The method may also be configured to reject the request in case the aggregated score is equal to the pre-determined threshold. Other decision logic may also be implemented.

[0042] According to various embodiments, a system 200 is disclosed configured to carry out the method as described herein in accordance with various embodiments. FIG. 9 shows a schematic system in accordance with various embodiments. The system 200 may an input interface 210 (e.g. a network interface) configured to receive the first financial data 142 and/or the benchmark financial data 110. The system 200 may further include a computer memory 204 configured to store the clusters. The system 200 may further include a processor 202, the processor 202 may be configured to carry out one or more of: the extracting and the segregating 120 of relevant data from the benchmark financial data into clusters; the analyzing 130 and generating a benchmark of at least one of the clusters; the determining a first credit score 144 of the entity based on the first financial data; the calculating an aggregated credit score 150 based on the first credit score and the benchmark.

[0043] The system 200 may include, for example, one or more servers, workstations, personal computers, laptops, tablets, smartphones, POS terminals, other suitable computing devices, etc. In addition, the system 200 may include a single computing device, or it may include multiple computing devices located in close proximity, or multiple computing devices distributed over a geographic region, so long as the system is specifically configured to function as described herein.

[0044] As shown in FIG. 9, an exemplary system 200 may be disclosed for sharing assessing financial statements. The system 200 may be hosted on a computing platform 200 and may include the computer platform. In some embodiments, the system may include a web-based or cloud-based platform that may be accessed over a network by computing devices 200. Referring to FIG. 9, the exemplary computing device 200 may include a processor 202 and a memory 204 coupled to (and in communication with) the processor 202.

[0045] According to various embodiments, the system 200 may further include a user input interface 208; and a user output interface, such as a presentation unit 206.

[0046] The processor 202 may include one or more processing units (e.g., in a multi-core configuration, etc.) including, without limitation, a central processing unit (CPU), a microcontroller, a reduced instruction set computer (RISC) processor, an application specific integrated circuit (ASIC), a programmable logic device (PLD), a gate array, and/or any other circuit or processor capable of the functions described herein. The above examples are exemplary only, and are not intended to limit in any way the definition and/or meaning of the processor.

[0047] The memory 204, as described herein, may be one or more devices that permit data, instructions, etc., to be stored therein and retrieved therefrom. The memory 204 may include one or more computer-readable storage media, such as, without limitation, dynamic random access memory (DRAM), static random access memory (SRAM), read only memory (ROM), erasable programmable read only memory (EPROM), solid state devices, flash drives, CD- ROMs, thumb drives, floppy disks, tapes, hard disks, and/or any other type of volatile or nonvolatile physical or tangible computer-readable media. The memory 204, and/or data structures included therein, may be configured to store, without limitation, first financial data, benchmark financial data, pre-defined data, and/or other types of data and/or information suitable for use as described herein. Furthermore, in various embodiments, computerexecutable instructions may be stored in the memory 204 for execution by the processor 202 to cause the processor 202 to perform one or more of the functions described herein, such that the memory 204 may be a physical, tangible, and non-transitory computer readable storage media. It should be appreciated that the memory 204 may include a variety of different memories, each implemented in one or more of the functions or processes described herein.

[0048] The computing device 200 may include a presentation unit 206 (or output device or display device) that is coupled to (and is in communication with) the processor 202 (however, it should be appreciated that the computing device 200 could include output devices other than the presentation unit 206, etc.). The presentation unit 206 outputs information, either visually or audibly to a user of the computing device 200, for example, representing an entity to be assessed, a user associated with either the credit issuing network, or the issuer, individuals associated with other parts of the method 100, etc. It should be further appreciated that various interfaces may be displayed at computing device 200, and in particular at presentation unit 206, to display information, such as, for example, financial summaries, aggregated credit score, etc. The presentation unit 206 may include, without limitation, a liquid crystal display (LCD), a light-emitting diode (LED) display, an organic LED (OLED) display, an “electronic ink” display, etc. In some embodiments, presentation unit 206 includes multiple devices.

[0049] The computing device 200 may include an input device 208 that is configured to receive inputs from the user of the computing device 200 (i.e., user inputs) such as, for example, the to initiate the upload of financial data, decide whether a decision is to approve or to reject the credit, or others. The input device 208 may be coupled to (and is in communication with) the processor 202 and may include, for example, a keyboard, a pointing device, a mouse, a stylus, a touch sensitive panel (e.g., a touch pad or a touch screen, etc.), another computing device, and/or an audio input device. In various exemplary embodiments, a touch screen, such as that included in a tablet, a smartphone, or similar device, behaves as both a presentation unit 206 and an input device 208.

[0050] The illustrated computing device 200 may include a network interface 210 coupled to (and in communication with) the processor 202 and the memory 204. The network interface 210 may include, without limitation, a wired network adapter, a wireless network adapter, a mobile network adapter, or other device capable of communicating to one or more different networks, including the network. Further, in some exemplary embodiments, the computing device 200 includes the processor 202 and one or more network interfaces incorporated into or with the processor 202.

[0051] Various embodiments relate to a non-transitory computer-readable medium storing computer executable code comprising instructions for determining a credit score according to the method in accordance with various embodiments.

[0052] Various embodiments relate to a computer executable code comprising instructions for determining a credit score in accordance with various embodiments.

[0053] FIG. 10 shows an exemplary interface 312 for input of financial data associate with an entity to be assessed by a user, in accordance with various embodiments. The interface may be part of a graphic user interface 310, for example, implemented on a monitor. The interface may include a region for drag files over for upload. Alternatively or in addition, the interface may include a button for the user to select the files to upload. The interface may include a confirmation button 314, e.g. including the text “Submit” for submitting the uploaded documents to be processed by the method in accordance with various embodiments.

[0054] FIG. 11 shows on the left side, one page of a consolidated balance sheet 320 of “The Company” which is an entity to be assessed. This balance sheet may be uploaded into the system together with other financial information as first financial data. A summary data 330 is shown on the right side as “Extracted information” which the user may edit and/or confirm.

[0055] FIG. 12 shows an exemplary summary report with the credit decision in accordance with various embodiments, which may be displayed on an output interface to a user, for example on a monitor. The summary report includes the credit decision, in this example “Reject”, the aggregated credit score, e.g., 0.4, the threshold used, e.g., 0.7, and the credit score e.g., 430. Further details may be shown such as current scenario and future projections for the entity, e.g. “Company A” and for the benchmark in the same industry sector.

[0056] Balance sheets, cash flow, income statements, such as included in the first financial data or benchmark financial data may be processed with statistical or machine learning methods, for example NLP. According to some embodiments, dictionaries including the name of the features to be determined, e.g. classified, may be used for finding the required features and/or for training a neural network. [0057] According to some embodiments, the names of balance sheet features may be selected from Cash; Accounts receivable; Inventory; Prepaid expenses; Short-term investments; Total current assets; Long-term investments; Property, plant, and equipment; (Less accumulated depreciation); Intangible assets; Total fixed assets; Deferred income tax; Other; Total Other Assets; Total Assets; Accounts payable; Short-term loans; Income taxes payable; Accrued salaries and wages; Unearned revenue; Current portion of long-term debt; Total current liabilities; Long-term debt; Deferred income tax; Other; Total long-term liabilities; Owner's investment; Retained earnings; Other; Total owner's equity; Year ; Total Liabilities and Owner's Equity.

[0058] According to some embodiments, the names of cash flow features may be selected from Cash Flow; Operating Cash Flow; Net Earnings; Plus: Depreciation & Amortization; Less: Changes in Working Capital; Cash from Operations; Investing Cash Flow; Investments in Property & Equipment; Cash from Investing; Financing Cash Flow; Issuance (repayment) of debt; Issuance (repayment) of equity; Cash from Financing; Net Increase (decrease) in Cash; Opening Cash Balance; Closing Cash Balance.

[0059] According to some embodiments, the names of income statement features may be selected from Income Statement; Sales revenue;(Less sales returns and allowances); Service revenue; Interest revenue; Other revenue; Total Revenues; Advertising; Bad debt; Commissions; Cost of goods sold; Depreciation; Employee benefits; Furniture and equipment; Insurance; Interest expense; Maintenance and repairs; Office supplies; Payroll taxes; Rent; Research and development; Salaries and wages; Software; Travel; Utilities; Web hosting and domains; Other; Total Expenses; Earnings Before Interest & Taxes; Interest Expense; Earnings Before Taxes; Income Taxes; Net Earnings.

[0060] According to some embodiments, the names of benchmark features may be selected from Category; Total current assets; Total fixed assets; Total Other Assets; Total Assets; Total current liabilities; Total long-term liabilities; Total owner's equity; Cash from Operations; Cash from Investing; Cash from Financing; Closing Cash Balance; Total Revenues; Total Expenses; Earnings Before Interest & Taxes; Interest Expense; Earnings Before Taxes; Income Taxes; Net Earnings.

[0061] FIG. 13 shows an exemplary flowchart of a method 400 for determining a credit score in accordance with various embodiments. In a step 402, benchmark financial data associated with benchmark entities is acquired, the benchmark financial data is anonymous data. In a step 404, a benchmark (also named benchmark score) is created by segment. In a step 406, first financial data associate with an entity to be assessed is acquired, including, e.g. loan details and Cash Flow information 406. In a step 408, a first credit score of the entity is determined based on the first financial data, for example, the first credit score may be received and/or may be computed. In a step 410, an aggregated credit score is calculated based on the first credit score and the benchmark. In a step 412 the aggregated credit score may be compared with a threshold for decision making. The disclosure is not limited thereto, and the method steps may be carried out in a different order, for example step 406 may be performed before step 402. Also, the method may include more or less method steps.

[0062] Credit analysts need to take into consideration past performance of similar underbanked customers which have grown through similar stages to determine if a new credit requestor will be obliging the repayments. Hence a holistic view is needed which can compare the past performance, present situation, and future movements to accurately carry out the credit risk assessment. In the current disclosure, an autonomous analysis of financial statements of various small medium enterprises is shown. The financial statement data is extracted and classified into various categories, e.g., related to retail, food and beverages, transportation, agriculture, chemical industry, construction, education, financial services, a combination thereof, or others. Benchmark scores of financial ratios may be created for comparison of segment merchant’s performance to determine credit worthiness of underbanked SMEs. An aggregated credit score is computed, which may be based on variable weightage to different benchmark score, credit score data if available and future projection of financial statement scores. The aggregated credit score may be normalized and checked against a threshold of credit worthiness determination to assess the credit risk and decision making.

[0063] While the disclosure has been particularly shown and described with reference to specific embodiments, it should be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. The scope of the invention is thus indicated by the appended claims and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced.

Claims

1. A method (100) for determining a credit score associate with an entity, the method being carried out by at least one microprocessor, and comprising: acquiring first financial data (142) associate with an entity to be assessed; acquiring benchmark financial data (110) associated with benchmark entities; extracting and segregating (120) relevant data from the benchmark financial data into clusters; analyzing (130) and generating a benchmark of at least one of the clusters; determining a first credit score (144) of the entity based on the first financial data; and calculating an aggregated credit score (150) based on the first credit score and the benchmark.

2. The method (100) of claim 1, further comprising a step of data anonymization wherein private data from the financial data is not carried over to the relevant data, wherein private data is identified according to pre-determined private data identification criteria.

3. The method (100) of claim 1 or claim 2, wherein extracting relevant data from the benchmark financial data comprises dark data analysis.

4. The method (100) of claim 1, wherein acquiring benchmark financial data (110) comprises acquiring anonymous benchmark financial data.

5. The method (100) of any one of the previous claims, wherein extracting relevant data from the benchmark financial data includes natural language processing (NLP) by a trained NLP processor.

6. The method (100) of any one of the previous claims, wherein segregating the relevant data into clusters is carried out by a trained classifier.

7. The method (100) of claim 6, wherein the trained classifier is configured to classify the relevant data into cluster according to a set of classes, the set of classes including industry sector classes including one or more of: retail, food, transportation, agriculture, chemical industry, construction, education, financial services.

8. The method (100) of any one of the previous claims, wherein analyzing (130) at least one of the clusters and generating a benchmark of at least one of the clusters is carried out for time frames of different duration.

9. The method (100) of any one of the previous claims, wherein determining a first credit score (144) comprises generating an A. I. score by a trained neural network.

10. The method (100) of any one of the previous claims, wherein determining a first credit score (144) comprises generating a traditional score by a deterministic score determination circuit.

11. The method (100) of any one of the previous claims, wherein calculating an aggregated credit score (150) comprises aggregating the benchmark with the first credit score, wherein the first credit score is weighted by a first pre-determined coefficient and the benchmark is weighted by a second pre-determined coefficient.

12. The method (100) of claim 10 or claim 11 wherein analyzing (130) and generating a benchmark of each of the clusters comprises compiling a benchmark summary data for each of the clusters and generating the benchmark from the benchmark summary data.

13. A system (200) configured to carry out the method of any one of claims 1 to 12, comprising: an input interface (210) configured to receive the first financial data (142) and the benchmark financial data (110); a computer memory (204) configured to store the clusters; a processor (202) configured to carry out: the extracting and the segregating (120) of relevant data from the benchmark financial data into clusters; the analyzing (130) and generating a benchmark of at least one of the clusters; 19 the determining a first credit score (144) of the entity based on the first financial data; and the calculating an aggregated credit score (150) based on the first credit score and the benchmark.

14. The system of claim 13 further comprising a user input interface (208); and a user output interface, such as a presentation unit (206).

15. The system of claim 13 or claim 14, wherein the benchmark financial data is anonymized.

16. A non-transitory computer-readable medium storing computer executable code comprising instructions for determining a credit score according to the method of any one of claims 1 to 13.

17. A computer executable code comprising instructions for determining a credit score according to the method of any one of claims 1 to 13.