System for financial analysis based on artificial intelligence with easy report interpretations

Abstract

Disclosed is a system for financial analysis for easy interpretation and recommendations The system 100 receives data through various input modules such as ERP, IOT devices, public data adopters, database management systems, financial statement or the like by a user 202 through a user interface unit 202. The system includes a processing unit 210 and a database unit 238 for storing and processing the financial and non-financial data. The processing unit 210 includes controller 236 and artificial intelligence module 240 for prefilling the input as per the trained data set stored in the database. The processing unit includes a data recommendation module 244 for financial statement analysis to provide recommendation reports for the user in the global languages. The system provides output in graphical and audible format in an interactive manner.

Description

[0001] “SYSTEM FOR FINANCIAL ANALYSIS BASED ON ARTIFICIAL

[0002] INTELLIGENCE WITH EASY REPORT INTERPRETATIONS”

[0003] FIELD OF THE INVENTION:

[0004] The present invention relates to data analysis, automation, data visualisation, and more particularly to artificial intelligence for financial and non-financial analysis with easy report interpretation.

[0005] BACKGROUND OF THE INVENTION:

[0006] Financial Management allows a better understanding of company’s performance, profit and strength. Financial management includes analysis of financial and non-financial statements of the company. Companies use online financial systems instead of manual labor. Online Financial system is becoming more popular than that of manual labor to process financial data and non-financial of the business. Manually entered financial system may have errors due to human factors such as carelessness.

[0007] Presently, many businesses and enterprises, for the purpose of budgeting, segment analysis, valuation, and similar activities heavily rely on tools such as excel or various Enterprise resource planning (ERPs). Organizations use software to manage day-to-day business activities such as accounting, procurement, project management, risk management and compliance, and supply chain operations. However, none of these software tools offer interpretation of the performed analysis.

[0008] Similarly, there are tools that can create prospective financial statements. However, the input mechanism is rudimentary such as filling data in excel file. There have been attempts in the prior art to overcome such problems some of them are discussed below.

[0009] Chinese Patent Application CN112287828A to CHEN XIAOLU relates to method and device for generating financial statements based on machine learning. The method includes steps as image identification processing, processing the invoice information, checking the invoice information, and judging whether the invoice information is correct or not. Further this invention teaches us generation of a financial statement according to the financial statement template obtained after copying.

[0010] Chinese Patent Application CN114565470A to WANG CHENXI relates to post-loan risk management and control of people. The method includes preprocessing stored financial data, data reclassification, data merging. Further feature extraction on preprocessed data extracts the features from the data. After performing data assessment final approval list is prepared. The method and device teach us only data analysis and generating final output list.

[0011] Korean Patent Application KR20210137596Ato Youngho Jeong is a method for financial analysis based on natural language processing. The system includes a big data collection unit that collects data received or generated in real time from media companies. The Korean Patent application includes a text pre-processing unit that analyzes the HTML structure of the data collected by the big data collection unit and parses only necessary information. Further, the system includes a natural language processing unit that performs morphological analysis on the original text of the parsed data to remove stop words and divides the document into word levels. The document morphologically analyzed by the natural language processing unit includes a price analysis unit for financial products by using a Deep Learning (Character Level Convolutional Neural Network: CL-CNN) prediction model and rules in units of words, and the word unit analysis unit is applied to individual financial products. The method is used for predicting the price of a financial product by collecting all information in real time. However, the easy prediction from visual report corresponding to financial statement is not taught in the prior art.

[0012] Hence, there is a need of a system for financial data analysis and its interpretation based on artificial intelligence for easy report interpretations. There is also a need of easy receiving of input from the user through web-based application with intuitive gaming experience techniques as through touch screen and mouse adaptive levers. There is also a need of generating the output of numerical financial analysis in global languages for ease of interpretation and creating insights.

[0013] SUMMARY OF THE INVENTION:

[0014] A system for financial analysis based on artificial intelligence with easy report interpretations include an user interface unit for receiving input data from users, a processing unit for processing received data and a database unit for storing processed data. The processing unit including a controller being configured to transmit and receive the data between the user interface and the database unit.

[0015] The user interface unit including a first data adaptor integrated with the ERP systems, a second data adaptor integrated with the IOT devices, a third data adaptor configured with the public databases, and a fourth data adaptor configured with the database management systems for receiving inputs. The user interface unit including a data importer module for receiving data through input forms and data file formats. The processing unit including a data collector module configured for aggregating and validating the influx of received data.

[0016] The processing unit including a data normalizer module configured for cleansing, enriching, transforming and standardizing the received data from the data collector module. The processing unit including a data formatter module storing and categorizing the processed data received from the data normalizer module. The processing unit including a data transformation module converting raw data into features. The processing unit including an artificial intelligence module prefilling the input, transforming system’s data into actionable intelligence and predicting financial risks on the data being received from the data formatter module.

[0017] The processing unit including a natural language processing module being configured for analyzing and extracting textual data from the user input. The processing unit including a data visualization module transforming complex dataset into insightful visualization. The processing unit including a data recommender module configured for processing the interpreted data for interpreting the output according to the user preferences.

[0018] The user interface unit including a login module for receiving user details, an authentication module for authenticating the registered users and a registration module for registering a new user. The first data adaptor configured for generating API calls and receiving data at preferably 1 millisecond. The first data adaptor is configured for correlating the data, imagining the relationship between two or more data points and accordingly structuring data to make it meaningful. The second data adaptor transforming the data in tranches, filtering, aggregating, and formatting conversion at the edge reducing the volume of data being transmitted to the system 100 and optimizing network bandwidth and storage utilization.

[0019] A method for parsing and transforming data in the third adaptor comprising steps parsing the data through regular expressions for identifying and extracting specific data patterns within text or code; parsing the data based on delimiters and on splitting data based on pre- separators; parsing, the data identifying and extracting elements based on tags and attributes; transforming, data from one data type to another; normalizing, the data representation across different sources, handling missing values, converting units, or applying specific data cleaning rules; and mapping extracted data elements to the system's internal data schema.

[0020] A method for enhancing data retrieval speed and performance in the fourth data adaptor comprising steps of creating, indexes on frequently accessed fields or columns for realtime data accessing; partitioning, involves dividing large datasets into smaller, more manageable units for reducing processing time and overall efficiency; and inmemory processing, significantly reducing retrieval times for commonly used data compared to accessing data stored on slower storage devices like hard drives.

[0021] The data importer module including validation rules being implemented for realtime assessment on the data entered, verifying its accuracy and completeness, the validation rules including assessment of data format, range validations, mandatory field completions, and cross-field dependencies.

[0022] The data importer module being configured with data cleansing operations such as normalization, duplication, and enrichment ensuring the quality and consistency of the imported data and aligning it with the system data standards and structures. The data normalizer module normalizes the data, by aligning different data formats, scales, and representations to a common standard including standardizing date formats, normalizing numerical values like currency or units of measurement, and converting text to a uniform case.

[0023] The data normalizer module including data enrichment feature to enhance the data by filling in missing information, adding context, or creating new derived attributes. The data normalizer module including data transformation that aggregates, sorts, and filters operations to prepare the data for targeted analysis. The artificial intelligence module transforming system's data into actionable intelligence, driving informed decision-making and strategic initiatives.

[0024] A method for extracting meaningful insights from the data through the artificial intelligence module comprising steps of preparing, cleaning, formatting, and preprocessing data to ensure it's suitable for analysis by the ML algorithms; selecting, machine learning algorithm and training on large dataset; evaluating, trained model’s performance on a separate dataset; integrating and deploying trained model to analyse new data and extract insights; predicting, recommending and automating decision based on extracted insights. A method for transforming complex datasets into clear and insightful visualizations through the data visualization module comprising steps of: selecting and filtering data, data subsets relevant to analysis needs; aggregating and summarizing large datasets for simplified visualization; mapping, the selected data elements to appropriate visual elements; assigning and mapping data values to visualized data; integrating interactive features to visualized data module; and transforming the data from steps a to gain clear and actionable insights from complex datasets.

[0025] The natural language processing module is configured to analyze text that involves processing and analyzing large volumes of textual data. A method for converting raw data into features through the data transformation module comprising steps of: gathering, domain specific data and utilizing statistical tools for identifying pattern, relationships and anomalies in the gathered data; cleansing data, by implementing, imputation techniques, identifying outliers and filtering noise to improve data quality and consistency; creating features, through mathematical operations and adding polynomial terms to capture non-linear relationships in data.

[0026] Converting continuous variables into categorical features and extracting cyclical features; transforming features, by normalizing, scaling, logarithmic transformations and handling non-normal data distribution; transforming, categorical data and text data into numerical form; selecting feature, by utilizing statistical features, machine learning algorithms or embedding within the model training process; reducing dimensionality, through the principal component analysis unit, distributed stochastic neighbor embedding (t-sne) processor and linear discriminant analysis (LDA) engine; training and evaluating, machine learning models iteratively and analyzing importance of features to optimize the performance; integrating tools for automation of parts of feature engineering process; and iterating above mentioned process steps for continuous refinement and improvement based on performance metrices.

[0027] BRIEF DESCRIPTION OF DRAWINGS:

[0028] The objectives and advantages of the present invention will become apparent from the following description read in accordance with the accompanying drawings wherein, FIG. 1 shows a system for financial analysis based on artificial intelligence for easy report interpretations in accordance with the present invention;

[0029] FIG.2 shows a system architecture of the system for financial analysis based on artificial intelligence for easy report interpretations of FIG.1; and

[0030] FIG. 3 shows a stepwise execution of the system for financial analysis based on artificial intelligence for easy report interpretations of FIG.1.

[0031] DESCRIPTION OF THE INVENTION:

[0032] References in the specification to "one embodiment" or "an embodiment" means that a particular feature, structure, characteristic, or function described in connection with the embodiment is included in at least one embodiment of the invention. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.

[0033] References in the specification to “preferred embodiment” means that a particular feature, structure, characteristic, or function described in detail thereby omitting known constructions and functions for clear description of the present invention.

[0034] The foregoing description of specific embodiments of the present invention has been presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the present invention to the precise forms disclosed and obviously many modifications and variations are possible in light of the above teaching.

[0035] The present invention is illustrated with reference to the accompanying drawings, throughout which reference numbers indicate corresponding parts in the various figures.

[0036] Referring to FIG. 1, a system for financial analysis based on artificial intelligence for easy report interpretation 100 (hereinafter referred to as “the system”) in accordance with a preferred embodiment of the present invention is described. The system 100 receives data input from various sources where the system 100 analyzes the received input and extrapolate meaningful patterns and insights, which are then distilled into reports of the characteristics of entered data. The system 100 is configured to receive input through various data adopters such as various ERP systems, various IOT devices, connection to stock exchanges, data management systems SQL, MongoDB or the like.

[0037] The system 100 also generates financial projections required for financial planning by analysis through visual and audible platform. The system 100 provides graphical visualization of financial projection. Accordingly, the system 100 also transmits the graphical visualization for display in an interactive manner. According to financial statement analysis the system 100 also provides the financial recommendations for the user in global languages.

[0038] A user 106 access the system 100 through a plurality of electronic devices 114. In this preferred embodiment the plurality of the electronic devices 114 are preferably Laptops, Computers, Handheld devices and the like however, the type of electronic devices varies in other embodiments of the present invention. The users 106 accesses the system 100 through communication network 110. In this preferred embodiment the communication network 110 is preferably Internet 110.

[0039] Referring to FIG. 2 the system 100 is described hereinafter. The system 100 includes a user interface unit 202, a processing unit 210 and a database unit 238. The user interface unit 202 includes a login module 204, an authentication module 208, a registration module 216 and a selector module 220. The user interface unit 202 also includes a first data adaptor 269, a second data adaptor 270, a third data adaptor 271, and a fourth data adaptor 272. In context of the present invention, the data adaptors 1st to 4thare configured to receive data through various data resources for example ERP, various IOT devices, stock exchanges, various databases or the like. However, system 100 is not limited to receiving data through the data adaptors from 1stto 4thand the data adaptors may vary in other embodiments of the present invention.

[0040] The processing unit 210 includes a first input processing module 228, a second input processing module 232, a data collector module 274, a data normalizer module 275, and a data formatter module 279. The processing unit 210 also includes a report generator module 277 and a data visualization module 268. Further, the processing unit 210 also includes a controller 236, an artificial intelligence module 240, a natural language processing module 234 and a data recommender module 244. The processing unit 210 also includes a data transformation module 281 configured to enhance the performance of machine learning algorithms by transforming raw input data received from the data adaptors into informative features. The user interface unit 202 stores and retrieves data from the database unit 238 through the processing unit 210.

[0041] The users 106 need to be registered with the system 100. The registration module 216 is configured for registration of a new user 106 in the system 100. After registration the users 106 enter the appropriate credentials and login to the system 100 through the login module 204. Further, the authentication module 208 validates the received credentials against the saved credentials in the database 238 and accordingly allows the registered users 106 to access the system 100. In this preferred embodiment, the input is received preferably from the first data adaptor 269, the second data adaptor 270, the third data adaptor 271, and the fourth data adaptor 272 through which financial data is actually entered in the system 100. In context of the present invention, the input is also received through input forms via a data importer module 273.

[0042] The first data adaptor 269 connects and collects data from various Enterprise Resource Planning (ERP) systems like Salesforce, SAP, Oracle, Tally ERP, or the like. The first data adaptor 269 is specialized for various ERP systems and includes specialized components designed to facilitate seamless integration with a wide range of Enterprise Resource Planning (ERP) systems. The first data adaptors 269 extracts vital organizational data locked within various ERP platforms and ensures its availability for comprehensive analysis within the system 100. In the first data adaptor 269 each ERP systems, related calls are made, these calls are often referred as API (Application Program Interface) calls. ERP system push data regularly into the system 100. The system 100 is also configurable to pull data from ERP at predefined time intervals preferably 1 millisecond.

[0043] The first data adaptor 269 is configured to convert ERP-specific data formats into a standardized format that is compatible with the system 100. The first data adaptor 269 includes a data cleaning unit that is a part of the overall process of data transformation. The first data adaptor 269 includes understanding the data, imagining the relationship between two or more data points and accordingly structuring it to make it meaningful. Similarly, there are possibilities of having null values or error values that are either deleted or computed on the basis of a use case.

[0044] The first data adaptor 269 also include robust error -handling and logging capabilities, essential for maintaining data integrity and facilitating troubleshooting. For example, while pulling data if error occurs and only partial data was pulled then the system 100 has capabilities to maintain the logs and once connection is reestablished, the incremental data is pulled. Accordingly, the first data adaptor 269 is for various ERP Systems and is a key enabler in the proposed system, allowing for the integration of comprehensive organizational data and laying the foundation for advanced analytics and business intelligence capabilities.

[0045] In accordance with the present invention, the second adaptor 270 is configured to interface with Internet of Things (loT) devices and sensors for collecting real-time data. The second data adaptor 270 captures, interprets, and transforms the data generated by loT devices into a format that is suitable for processing and analysis within the system 100. The second data adaptor 270 communicates with a wide range of loT devices, each potentially using different communication protocols and data formats. The second data adaptor 270 receives data in real-time as the loT devices often generate data continuously and in large volumes, hence, the second data adaptor 270 is adaptable for high throughput and low latency. The second data adaptor 270 utilizes Json format. Consequently, by coupling the given format with high computing processors, the second data adaptor 270 achieve high throughput and low latency. The second data adaptor 270 ensures that the data is processed in tranches to avoid unnecessary load onto system 100. The second adaptor 270 is configured to perform initial data processing tasks such as filtering, aggregation, and format conversion at the edge thus reducing the volume of data transmitted to the system 100 and optimizing network bandwidth and storage utilization.

[0046] The third data adaptor 271, is configured to integrate with publicly available data sources such as stock exchanges, company reports, and government -run websites and databases. The third data adaptor 271 captures a broad spectrum of data, including financial market trends, demographic statistics, environmental data, and social media insights, thereby providing a more comprehensive analytical landscape. The third data adaptor 271 receives data from data sources through leveraging web scraping technologies to extract information from websites, utilizes APIs to connect with public databases and online services, and employs RSS feeds for real-time news and updates.

[0047] The third data adaptor 271 includes advanced data parsing and transformation techniques to normalize and format the extracted data, making it compatible with the data schema of the system 100. This ensures seamless integration and consistency across different data sets. The third data adopter 271 is configured to parse data by entering a regular expression for identifying and extracting specific data patterns within text or code that is useful for structured data with consistent delimiters like commas or tabs.

[0048] Further, the third data adaptor 271 is configured with delimiter-based parsing that focuses on splitting data based on pre -defined separators. The third data adaptor 271 is also configured with HTML / XML parsing for structured data stored in HTML or XML formats. The HTML parsers identify, and extract elements based on tags and attributes.

[0049] The third data adaptor 271 includes data transformation that has a data type conversion that transforms data from one format to another e.g., string to integer, date formatting. The third data adaptor 271 also normalizes the received data ensuring consistent data representation across different sources this involves handling missing values, converting units, or applying specific data cleaning rules. The third data adaptor 271 includes a schema mapping that maps extracted data elements to the system's 100 internal data schema that ensures the data is stored and read in a consistent manner within the system 100.

[0050] The fourth data adaptor 272 receives and stores data using systems like SQL, MongoDB, Firestore, Oracle, Big Query or the like. In the fourth adaptor 272 a combination of relational databases, data warehouses, and data lakes are data sources. Relational databases are utilized for managing transactional data, offering fast and efficient read and write capabilities, which are essential for real-time data processing and operational reporting. The fourth data adaptor 272 is configured to handle structured data with complex relationships, ensuring data integrity and consistency through ACID (Atomicity, Consistency, Isolation, Durability) properties. The fourth data adaptor 272 includes features such as indexing, partitioning, and in-memory processing to enhance data retrieval speeds and performance.

[0051] In the fourth data adaptor 272 the data is indexed that, creates a structured reference system for data storage. Indexing on frequently accessed fields or columns, increases data retrieval speeds. This indexing is crucial for real-time or near real-time data access needs in fourth data adaptor 272. In context of the present invention, the fourth data adaptor 272 partitions the data that involves dividing large datasets into smaller, more manageable units. Data partitioning enhances query performance by allowing the system 100 to focus on the relevant data subset for each search that reduces processing time and improves overall efficiency.

[0052] In the present invention, the fourth data adaptor 272 includes an in-memory processing technique that keeps frequently accessed data readily available in the system 100 main memory (RAM) for faster processing. The in-memory processing significantly reduces retrieval times for commonly used data compared to accessing data stored on slower storage devices like hard drives.

[0053] The data importer module 273 includes receiving data through the input forms to collect data directly from users 105. The data importer module 273 has an architecture that focuses around a user-friendly interface, allowing users from various backgrounds to input data efficiently and accurately. The data importer module 273 is equipped with intuitive fields, drop-down menus, checkboxes, and date selectors to accommodate diverse data types and formats. The data importer module 273 is structured to guide users through the data entry process, minimizing the likelihood of errors and ensuring data consistency.

[0054] The data importer module 273 includes advanced validation rules these rules are implemented to perform real-time checks on the data entered, verifying its accuracy and completeness. The validation rules includes checks for data format, range validations, mandatory field completions, and cross-field dependencies. The data importer module 273 validates data at the point of entry, that reduces the need for subsequent data cleaning and processing.

[0055] The data importer module 273 facilitate the import of large volumes of data into the system. The data importer module 273 is configured to efficiently handle the ingestion of large datasets from various sources. The data importer module 273 is configured to receive diverse range of file formats, such as PDF, Excel, CSV, JSON, HTML, and XML, catering to the varied data storage practices across different organizations. The data importer module 273 is configured to sophisticatedly parse mechanisms to accurately extract data from the received files. During the import process, the data importer module 273 performs crucial data cleansing operations such as normalization, duplication, and enrichment. These processes are essential to ensure the quality and consistency of the imported data, aligning it with the system 100 data standards and structures.

[0056] The data importer module 273 also includes error handling that robustly identifies and logs any inconsistencies or issues encountered during the import process, such as format mismatches, missing mandatory fields, or data type errors. This feature allows users to address and rectify these issues promptly, ensuring data integrity. The data importer module 273 is equipped with tools for monitoring and managing the import process, providing users with visibility into the progress, status, and any issues encountered during data ingestion. This transparency is crucial for maintaining control and confidence in the data import process.

[0057] The selector module 220 is configured to receive the input. In accordance with the present invention, the selector module 220 is configured to provide list of choices for receiving data to the system 100. The user 106 selects at least one input method for receiving data to the system 100.

[0058] Further, the first input processing module 228 is configured on the controller 236 for processing and analyzing the data received from the data adaptor 269, 270, 271 , 272 in the system 100. The analyzed data from the first input processing module 228 is then transmitted to the output unit 230 and the recommendation module 244. The second input processing module 232 is configured on the controller 236 for processing and analyzing the data received from the data importer module 273. The first input processing module 228 is configured to process and analyze the provided data.

[0059] The data collector module 274 aggregates data from the first to fourth adaptors 269, 270, 271 and 272 and the data importer module 273. The data collector module 274 aggregates and manages the influx of data from a multitude of sources. The data collector module 274 receives various data formats, ranging from structured formats like CSV and JSON to unstructured formats like logs and raw sensor outputs. The data collector module 274 includes robust validation mechanisms to ensure data integrity from the outset. The data collector module 274 checks for data completeness, format consistency, and adherence to predefined schemas, thereby filtering out erroneous or irrelevant data at the entry point.

[0060] Additionally, the data collector module 274 is highly scalable, accommodates fluctuating data volumes without compromising on performance. This scalability is crucial in environments with variable data inflows, such as those experiencing periodic spikes in data generation. The data collector module 274 also has time -stamping functionalities, that maintain data lineage and enable time -based analyses.

[0061] The analyzed data from the second input processing module 232 and data collector module 274 is then transmitted to the data normalizer module 275. The data normalizer module 275 converts the diverse and disparate data collected from various sources into a uniform, standardized format. The data normalizer module 275 normalizes the data, by aligning different data formats, scales, and representations to a common standard. Data normalization includes standardizing date formats, normalizing numerical values like currency or units of measurement, and converting text to a uniform case. Normalization is vital for enabling seamless data integration and comparison across different datasets.

[0062] Further, the data normalizer module 275 is configured for extensive data cleaning such as identifying and correcting errors and inconsistencies in the data, missing values, duplicates, or outliers. The data normalizer module 275 includes pattern recognition, rule-based algorithms, and statistical methods that detect and rectify said anomalies to maintain the integrity and reliability of the data.

[0063] The data normalizer module 275 includes data enrichment feature to enhance the data by filling in missing information, adding context, or creating new derived attributes. Data enrichment also integrates external data sources, calculates new metrics, or applies business rules to add value to the existing data. The data normalizer module 275 transforms the data that reshapes and restructures the data as needed for specific analytical requirements. The data transformation includes aggregations, sorting, and filtering operations to prepare the data for targeted analysis. The data formatter module 279 stores processed data ready for analysis. The data formatter module 279 includes structured storage and categorization of processed data, distinguishing between financially and non-financially significant information. The data formatter module 279 module intricately stores a wide array of data types. The data formatter module 279 module methodically categorizes and stores financial data, including but not limited to sales figures, budget reports, and market data, facilitating easy access and analysis. Concurrently, non-financial data such as customer feedback, employee performance metrics, and operational data are also systematically organized and stored. This dual categorization allows for targeted analysis and reporting, enabling users to focus on either financial or non-financial aspects as needed.

[0064] The data transformation module 281 is an integral part of the machine learning pipeline that converts raw data into features that are optimized for use in machine learning algorithms. The data transformation module 281 includes multiple stages i.e., data understanding and exploration, data cleaning, feature creation, feature transformation, feature encoding, feature selection, dimensionality reduction, and iterative evaluation for improving the predictive accuracy of machine learning models.

[0065] In context of the present invention, the data understanding and exploration includes a domain analysis unit and an exploratory data analysis unit. Further, the data cleaning unit includes a value handler, an outlier detection and treatment system and a noise reduction mechanism. The feature creation module includes an interaction feature generator, a polynomial feature constructor, a binning processor and a datetime feature extractor. The feature transformation module includes a normalization / scaling unit, a log transformation engine and an advanced transformation techniques.

[0066] The feature encoding unit includes a categorical encoder and a text feature extraction module. The feature selection module includes a filter method, wrapper methods and embedded method. The dimensionality reduction module includes a principal component analysis unit, linear discriminant analysis engine. The evaluation module includes model training and evaluation framework and feature importance analyzer. The automation module includes an automated feature tool.

[0067] In accordance with the present invention, the domain analysis is configured to gather domain specific knowledge to understand the problem context and data characteristics. The exploratory data analysis (EDA) toolset that is configured to utilize statistical and visualization tools for identifying patterns, relationships, and anomalies within the gathered data. Further, the value handler implements techniques such as imputation (mean, median, mode) or deletion to address missing data. Furthermore, an outlier detection and treatment system identifies and processes outliers through methods like clipping, transformation, or removal. Further, the noise reduction mechanism filters out noise to improve data quality and consistency.

[0068] Further, the interaction feature generator creates new features by combining existing ones through mathematical operations such as addition, multiplication, and ratios. The polynomial feature constructor is configured to add polynomial terms to capture non-linear relationships in the data. Moreover, the binning processor converts continuous variables into categorical features by segmenting the range into intervals. The datetime feature extractor is configured to extract components from datetime data, such as day of the week, month, year, or cyclical features.

[0069] Accordingly, the normalization / scaling unit is configured with min-max scaling, standard scaling, or other normalization techniques to bring features to a similar scale. The log transformation engine is configured to reduces skewness and stabilize variance through logarithmic transformations. The advanced transformation techniques are configured to implement box-cox or yeo-johnson transformations for handling nonnormal data distributions. Further, the categorical encoder is configured to convert categorical data into numerical form using methods such as one -hot encoding, label encoding, target encoding, or frequency encoding. Moreover, the text feature extraction module is configured to transform text data into numerical features using techniques like TF-IDF, word embeddings (Word2Vec, GloVe), or deep learning methods. Further, the filter methods utilize statistical techniques (e.g., correlation coefficients, Chi-square tests) to select relevant features. The wrapper methods employ machine learning algorithms (e.g., recursive feature elimination) to evaluate and select feature subsets. Further, the embedded methods are incorporated with feature selection within the model training process using techniques like LASSO and Ridge regression.

[0070] Moreover, the principal component analysis (PCA) unit reduces dimensionality by transforming features into a set of linearly uncorrelated components. The distributed stochastic neighbour embedding (T-SNE) processor visualizes high-dimensional data by reducing it to two or three dimensions. The linear discriminant analysis (LDA) engine is configured to extract features by finding linear combinations that best separate different classes.

[0071] The model training and evaluation framework is configured to train and evaluate models iteratively using engineered features to optimize performance. Further, the feature importance analyzer is configured to assess the importance of features and refine the feature engineering process based on these insights. The automated feature engineering tools integration incorporates tools and libraries (e.g., FeatureTools, TPOT, AutoME platforms) to automate parts of the feature engineering process. Further, the custom pipelines constructor creates reusable and scalable pipelines using libraries such as scikit-learn and pandas, alongside custom scripts.

[0072] The artificial intelligence module 240 is configured for prefilling the input as per the trained data set stored in the database. Additionally, the artificial intelligence module 240 also allows the user to selectively increase or decrease volume of the data. The artificial intelligence module 240 is configured for analyzing large amount of received input data for e.g., analysis of liquidity, solvency, profitability position, cash flow balance sheets and the like. The artificial intelligence module 240 is configured for predicting changes in consumers demand, that adjusts pricing strategy and inventory levels in the business. The artificial intelligence module 240 transforms system's data into actionable intelligence, driving informed decision-making and strategic initiatives. The artificial intelligence module 240 is configured for leveraging both supervised and unsupervised learning algorithms. In the artificial intelligence module 240 predictive modeling is a key aspect of this service, where machine learning models are trained on historical data to forecast future trends, identify potential risks, and uncover opportunities.

[0073] The artificial intelligence module 240 detects anomaly such that the service utilizes statistical and machine learning techniques to identify unusual patterns or outliers in the data that is used for fraud detection, quality control, and network security. Initially the artificial intelligence module 240 includes data preparation in which data is cleaned, formatted, and pre-processed to ensure it's suitable for analysis by the ML algorithms. Further, the artificial intelligence module 240 selects machine learning algorithm and accordingly trains a large dataset. The artificial intelligence module 240 trains the dataset included for feeding the data into the algorithm and adjusting its internal parameters to improve its ability to recognize patterns.

[0074] Further the trained model's performance is evaluated on a separate dataset to assess its accuracy and generalizability to effectively identify patterns in new, unseen data. The artificial intelligence module 240 is deployed and it's integrated into the Al system. The artificial intelligence module 240 then uses the trained model to analyze new data and extract insights. Further, based on the extracted insights, the artificial intelligence module 240 is configured to make predictions, recommendations, or automated decisions.

[0075] The data visualization module 268 creates visual representations of data findings. The data visualization module 268 is a multi-step process to transform complex datasets into clear and insightful visualizations. The data visualization module 268 breaks down the data through selecting and filtering data, aggregating and summarization, mapping to visual elements, generating interactive features and visualizing the data to be displayed. Initially, the data to be visualized is selected and filtered that allows users 106 to select specific data subsets relevant to their analysis needs reducing information overload and enabling focus on the visualization of the key aspects of interest. The users 106 filter data based on various criteria, such as date range, specific categories, or custom filters.

[0076] Further, the data is aggregated, and the large datasets are overwhelmed to visualize directly. The visualization module 268 includes functionalities to aggregate and summarize data that includes summarizing data by calculating statistics like mean, median, or standard deviation. Grouping data into categories for visualization e.g., customer demographic or the like.

[0077] Moreover, the data is mapped to visual elements after the data is selected and prepared. The visualization module 268 maps the data attributes to appropriate visual elements within the chosen chart type that includes assigning data values to bar heights, line positions, or pie chart slice sizes, mapping data categories to colors or symbols within the visualization.

[0078] The data visualization module 268 incorporates interactive elements to enhance user experience and exploration. In the data visualization module 268 the users hover over data points to view detailed information tooltips. Further, zooming and panning functionalities allow users to focus on specific areas of the visualization. Moreover, interactive filters enable users to dynamically filter the data and update the visualization in real-time. In the data visualization module 268 the module is configured to create clear and aesthetically pleasing visualizations that includes using color palettes that are easy to distinguish and avoid color blindness issues. Further, the data visualization module 268 include applying appropriate chart types based on the data and the insights intended for communication. Further, the data visualization module 268 design implements clear and concise labelling of axes, chart elements, and legends.

[0079] The data visualization module 268 empowers users to gain clear and actionable insights from complex datasets by transforming data through said steps. The data visualization module 268 enables the users to explore the data interactively, identify patterns and trends, and effectively communicate their findings to others.

[0080] The data recommender module 244 is configured for receiving the analyzed and calculated data from the second input processing module 232 or from the first input processing module 228 or from the natural language processing module 234. The data recommender module 244 further processes the received data and provides predictions about market trends and consumer behavior as recommendation data to the output unit 230. The data recommender module 244 is configured to predict areas of improvements in the business. Also, the recommendation module 244 processes the interpreted data for interpreting the output according to customer preferences and requirements.

[0081] Additionally, the natural language processing module 234 is configured for processing and analyzing inputs such as customer feedback, social media posts. The natural language processing module 234 enables interaction with the system 100 using natural language queries. The natural language processing module 234 enables the system 100 to understand, interpret, and respond to text and voice inputs in a natural and intuitive manner, making the system 100 more accessible and user-friendly.

[0082] The natural language processing module 234 is configured to analyze text that involves processing and analysing large volumes of textual data. The text analysis include extracting key phrases, identifying themes, and sentiment analysis, that are invaluable for understanding customer feedback, social media sentiment, or market trends. The natural language processing module 234 is configured for hands-free data querying, report generation, and accessibility purposes. The voice recognition system is designed to accurately capture and transcribe spoken words, even in noisy environments or with various accents and dialects. The natural language processing module 234 includes language translation capabilities that enable the system 100 to operate across different languages, broadening its usability and appeal. The language translation automatically translate text and voice inputs into multiple languages, facilitating seamless communication and data analysis for users worldwide. In accordance with the present invention, the output unit 230 includes an audio output 264 [not shown] and a visual output 268 [not shown]. In this preferred embodiment, the audio output 264 is speaker. The audio output provides interpretation output and recommendation output. Also, the visual output 268 is preferably display. The visual output 268 provides interpretation output and recommendation output as graphical representation.

[0083] The user interface unit 202 provides facility of selection of features and data output and accordingly automates audible recommendation in user friendly language. Additionally, the user interface 202 provides facility of selection of features and data output and accordingly generates visual output as interpretation and automates recommendation by user interactive graphical representation, models presentation of financial and non-financial data valuations.

[0084] The report generator module 277 generates reports and insights in multiple languages. The report generator module 277 is configured to convert all the information in that report into a variety of languages. The report generator module 277 ensures that the language used is clear, accurate, and appropriate for the content in which even complex data concepts and analytics are communicated in a way that is easy for everyone to grasp, whether they are data experts or not. In the report generator module 277 user selects their preferred language and the report is translated accordingly.

[0085] Now referring to FIGS. 1 to 3, an operational flow of the system for financial analysis based on artificial intelligence for easy report interpretations 100 is described. In an initial step 304, the users 106 access the system 100 through the electronic devices 114. In a next step 308, the users 106 get registered in the system by the registration module 216. In this step 308, the registered users 105 logs into the system 100 through the login module 204. In this step 308, the system validates the details by the authentication module 208.

[0086] In a further step 312, the user 106 is provided with input methods through the selector module 220. In this step 312, the user select the input method between the first data adaptor 269, the second data adaptor 270, the third data adaptor 271, the fourth data adaptor 272 and the data importer module 273. In this step 312, the data adaptors captures interprets and transforms the data received from the respective input mechanisms.

[0087] In a next step 316, the received data from the data adaptors and data importer module 269, 270, 271, 272 and 273 is sent to the first input processing module 228, the second input processing module 232 and the data collector module 274 by the controller 236 for processing the data. In this step 316, the data collector module 274 checks for completeness, format consistency and sends the towards step 320. In a next step 320, the data normalizer module 275 is configured to normalize and process the data for seamless data integration.

[0088] In this step 320, the financial and non-financial data is normalized and formatted and send towards step 324. In this step 324 the processed and formatted data is prefilled as per the trained data set stored in the database. In this step 324, the artificial intelligence module 240 selects machine learning algorithm and accordingly trains a large dataset. In a next step 328, the processed data is visualized that is a multi- step process to transform complex datasets into clear and insightful visualizations. In a next step 332, the data recommender module 244 is configured to predict areas of improvements in the business. In a next step 336, the report generator module 277 generates reports and insights in multiple language.

[0089] The system 100 advantageously, provides graphical interpretation of financial analysis of data that helps to identify potential risks. The system 100, suggest users the recommended actions that helps user for quick and correct decision making. The system 100

[0090] The embodiments were chosen and described in order to best explain the principles of the present invention and its practical application, to thereby enable others, skilled in the art to best utilize the present invention and various embodiments with various modifications as are suited to the particular use contemplated. It is understood that various omission and substitutions of equivalents are contemplated as circumstance may suggest or render expedient, but such are intended to cover the application or implementation without departing from the scope of the present invention.

Claims

CLAIMS:

1. A system for financial analysis based on artificial intelligence with easy report interpretations 100 comprising: an user interface unit 202 for receiving input data from users 106, a processing unit 210 for processing received data and a database unit 238 for storing processed data, the processing unit 210 including a controller 236 being configured to transmit and receive the data between the user interface 210 and the database unit 238; the user interface unit 202 including a first data adaptor 269 integrated with the ERP systems, a second data adaptor 270 integrated with the IOT devices, a third data adaptor 271 configured with the public databases, and a fourth data adaptor 272 configured with the database management systems for receiving inputs; the user interface unit 202 including a data importer module 273 for receiving data through input forms and data file formats; the processing unit 210 including a data collector module 274 configured for aggregating and validating the influx of received data; the processing unit 210 including a data normalizer module 275 configured for cleansing, enriching, transforming and standardizing the received data from the data collector module 274; the processing unit 210 including a data formatter module 279 storing and categorizing the processed data received from the data normalizer module 275;the processing unit 210 including a data transformation module 281 converting raw data into features; the processing unit 210 including an artificial intelligence module 240 prefilling the input, transforming system’s data into actionable intelligence and predicting financial risks on the data being received from the data formatter module 279; the processing unit 210 including a natural language processing module 234 being configured for analyzing and extracting textual data from the user input; the processing unit 210 including a data visualization module 268 transforming complex dataset into insightful visualization; and the processing unit 210 including a data recommender module 244 configured for processing the interpreted data for interpreting the output according to the user 106 preferences.

2. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the user interface unit 202 including a login module 204 for receiving user details, an authentication module 208 for authenticating the registered users and a registration module 216 for registering a new user 106.

3. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the first data adaptor 269 configured for generating API calls and receiving data at preferably 1 millisecond.

4. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the first data adaptor 269 is configured for correlating the data, imagining the relationship between two or more data points and accordingly structuring data to make it meaningful.

5. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the second data adaptor 270 transforming the data in tranches, filtering, aggregating, and formatting conversion at the edge reducing the volume of data being transmitted to the system 100 and optimizing network bandwidth and storage utilization.

6. A method for parsing and transforming data in the third adaptor 271 as claimed in claim 1 comprising steps of: a. parsing the data through regular expressions for identifying and extracting specific data patterns within text or code; b. parsing the data based on delimiters and on splitting data based on pre- separators; c. parsing, the data identifying and extracting elements based on tags and attributes; d. transforming, data from one data type to another; e. normalizing, the data representation across different sources, handling missing values, converting units, or applying specific data cleaning rules; andf. mapping extracted data elements to the system's 100 internal data schema.

7. A method for enhancing data retrieval speed and performance in the fourth data adaptor 272 as claimed in claim 1 comprising steps of: a. creating, indexes on frequently accessed fields or columns for realtime data accessing; b. partitioning, involves dividing large datasets into smaller, more manageable units for reducing processing time and overall efficiency; and c. in-memory processing, significantly reducing retrieval times for commonly used data compared to accessing data stored on slower storage devices like hard drives.

8. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the data importer module 273 including validation rules being implemented for real-time assessment on the data entered, verifying its accuracy and completeness, the validation rules including assessment of data format, range validations, mandatory field completions, and cross-field dependencies.

9. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the data importer module 273 being configured with data cleansing operations such as normalization, duplication, and enrichment ensuring the quality andconsistency of the imported data and aligning it with the system 100 data standards and structures.

10. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the data normalizer module 275 normalizes the data, by aligning different data formats, scales, and representations to a common standard including standardizing date formats, normalizing numerical values like currency or units of measurement, and converting text to a uniform case.

11. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the data normalizer module 275 including data enrichment feature to enhance the data by filling in missing information, adding context, or creating new derived attributes.

12. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the data normalizer module 275 including data transformation that aggregates, sorts, and filters operations to prepare the data for targeted analysis.

13. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the artificial intelligence module 240 transforming system's data into actionable intelligence, driving informed decision-making and strategic initiatives.

14. A method for extracting meaningful insights from the data through the artificial intelligence module 240 as claimed in claim 1 comprising steps of:a. preparing, cleaning, formatting, and pre-processing data to ensure it's suitable for analysis by the ML algorithms; b. selecting, machine learning algorithm and training on large dataset; c. evaluating, trained model’s performance on a separate dataset; d. integrating and deploying trained model to analyse new data and extract insights; and e. predicting, recommending and automating decision based on extracted insights.

15. A method for transforming complex datasets into clear and insightful visualizations through the data visualization module 268 as claimed in claim 1 comprising steps of: a. Selecting and filtering data, data subsets relevant to analysis needs; b. Aggregating and summarizing large datasets for simplified visualization; c. Mapping, the selected data elements to appropriate visual elements; d. Assigning and mapping data values to visualized data; e. integrating interactive features to visualized data module; and f. transforming the data from steps a to gain clear and actionable insights from complex datasets.

16. The system for financial analysis based on artificial intelligence with easy report interpretations 100 as claimed in claim 1 wherein, the natural languageprocessing module 234 is configured to analyze text that involves processing and analyzing large volumes of textual data.

17. A method for converting raw data into features through the data transformation module 281 as claimed in claim 1 comprising steps of: a. Gathering, domain specific data and utilizing statistical tools for identifying pattern, relationships and anomalies in the gathered data; b. Cleansing data, by implementing, imputation techniques, identifying outliers and filtering noise to improve data quality and consistency; c. Creating features, through mathematical operations and adding polynomial terms to capture non-linear relationships in data; d. Converting continuous variables into categorical features and extracting cyclical features; e. Transforming features, by normalizing, scaling, logarithmic transformations and handling non-normal data distribution; f. Transforming, categorical data and text data into numerical form; g. Selecting feature, by utilizing statistical features, machine learning algorithms or embedding within the model training process; h. Reducing dimensionality, through the principal component analysis unit, distributed stochastic neighbor embedding (t-sne) processor and linear discriminant analysis (LDA) engine ; i. Training and evaluating, machine learning models iteratively and analyzing importance of features to optimize the performance;j. Integrating tools for automation of parts of feature engineering process; and k. Iterating above mentioned process steps for continuous refinement and improvement based on performance metrices.

Images