A multi-factor weight AI analysis model and method based on linear regression

Abstract

The application relates to the field of artificial intelligence and data statistical analysis, in particular to a multi-factor weight AI analysis model and method based on linear regression, which is composed of a multi-source data preprocessing module, an intelligent modeling and weight quantification module and an automatic sorting and visual output module, the preprocessing module carries out missing value filling, abnormal value elimination and Z-score standardization; the modeling module constructs a multiple linear regression model with a target result as a dependent variable and various influence factors as independent variables, introduces L2 regularization optimization through K-fold cross validation to lock the optimal penalty coefficient, multiplies the regression coefficient by the ratio of the standard deviation of the independent variable and the dependent variable to obtain a standardized regression coefficient, and takes the absolute value as a real influence weight; the output module sorts in descending order according to the weight and generates an analysis report. The application integrates data cleaning, modeling, weight quantification, sorting and visualization, eliminates dimensional interference, solves collinearity distortion, objectively quantifies the contribution degree and reduces the analysis threshold.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and data statistical analysis technology, specifically to a multi-factor weighted AI analysis model and method based on linear regression. Background Technology

[0002] In scientific experiments, data analysis, and social surveys, research results are often influenced by multiple complex factors. Accurately distinguishing the primary and secondary relationships of these factors and quantifying their influence weights are crucial for optimizing experimental designs and drawing scientific conclusions. Currently, multi-factor influence analysis mainly faces the following challenges and limitations: First, traditional manual analysis is highly subjective. Traditional analysis methods rely heavily on researchers' human experience to judge the importance of factors, which is highly subjective, has low quantitative accuracy, and cannot be scientifically ranked, resulting in a cumbersome analysis process and poor reliability of conclusions. Secondly, professional software has a high barrier to entry. Although existing professional modeling and statistical software such as SPSS and R language have professional functions, they have a high operating threshold and high learning cost. They require users to have a solid foundation in statistics or programming, making it difficult to achieve lightweight and fast multi-factor influence analysis. They are not suitable for undergraduate scientific research and basic data analysis scenarios. Third, the tools are fragmented. Most similar tools only provide single modeling or statistical functions and lack a one-stop solution that integrates data cleaning, model training, weight quantification, intelligent sorting and visualization. This results in users having to switch between multiple software programs, making the process cumbersome.

[0003] Therefore, there is an urgent need for an intelligent analysis model that is lightweight and can automatically quantify the true impact weights of each factor and accurately rank them, in order to solve the pain points of ambiguous primary and secondary factors, subjective manual analysis, and difficulty in operating professional tools in multivariate studies. Summary of the Invention

[0004] To address the limitations of existing multi-factor data analysis techniques in basic scientific research and engineering practice, the technical problems to be solved by this invention include: First, traditional analysis suffers from high subjectivity and low quantitative accuracy. Existing multifactor analysis relies heavily on researchers' prior experience to determine primary and secondary factors, lacking objective quantitative standards. The partial regression coefficients of traditional multiple regression are affected by the dimension, making direct horizontal comparison impossible, which makes it difficult to accurately separate and measure the true contribution of factors. Second, complex multicollinearity leads to model distortion and inaccurate weights. In real multidimensional datasets, there are often hidden cross-collinearities among influencing factors, such as the correlation between drug dosage and administration time. When faced with collinearity, the regression coefficients of the traditional ordinary least squares method are prone to shift or even sign reversal, resulting in serious distortion of the calculated weights, which cannot reflect the independent physical contribution of the factors. Third, advanced statistical tools have high barriers to entry, fragmented processes, and a lack of adaptability. Existing professional statistical software such as SPSS and R language are complex to operate and have steep learning curves. Furthermore, data cleaning, modeling, weight calculation, sorting, and graphing are disconnected from each other, requiring frequent manual intervention and switching. At the same time, models with fixed parameters are difficult to adapt to datasets with different distribution characteristics, and are prone to underfitting or overfitting, lacking parameter optimization and adaptive adjustment mechanisms.

[0005] To address the aforementioned technical problems, this invention provides a multi-factor weighted AI analysis model and method based on linear regression, comprising the following steps: Step S1, Multi-source data integration and preprocessing: Obtain multi-dimensional raw data uploaded by users, automatically perform missing value filling, outlier removal and data normalization processing, eliminate differences in units, and standardize the input of multi-factor data; Step S2: Construct a multiple linear regression AI model: Using the target research outcome as the dependent variable and various influencing factors as independent variables, construct a multiple linear regression model. Iterate and fit the model using the least squares method combined with the AI ​​adaptive iterative optimization mechanism to establish the correlation between factors and results. Step S3, Intelligent Weight Automatic Calculation: Based on the model regression coefficients, standardization transformation is performed to quantify the true impact weight of each factor, and the contribution of all influencing factors is objectively quantified; Step S4: Automatic Priority Sorting of Influence Factors: Based on the calculated weight values, the influencing factors are automatically sorted in descending order from largest to smallest, distinguishing between core influencing factors and weak influencing factors. Step S5, Result Visualization and Data Output: Display the calculation results through charts, weight lists, and analysis reports, and export the data and generate reports.

[0006] Furthermore, in step S1, the data normalization process employs Min-Max standardization or Z-score standardization to map influence factors of different dimensions and orders of magnitude to a unified interval.

[0007] Furthermore, in step S2, the AI ​​adaptive iterative optimization mechanism is as follows: during the least squares fitting process, the model parameters are dynamically adjusted, a regularization term is introduced to prevent overfitting, and the optimal feature combination is adaptively selected based on the characteristics of the dataset.

[0008] Further, in step S3, the standardization transformation specifically involves multiplying the partial regression coefficients obtained from the multiple linear regression model by the ratio of the standard deviation of the corresponding independent variable to the standard deviation of the dependent variable, thereby converting them into standardized regression coefficients, and using the absolute value of the standardized regression coefficients as the true influence weight of each factor.

[0009] A multi-factor weighted AI analysis system based on linear regression includes a multi-source data preprocessing module, an intelligent modeling and weight quantification module, and an automatic sorting and visualization output module; The multi-source data preprocessing module acquires multi-dimensional raw data in batches and automatically completes missing value filling, outlier removal, and data normalization. The intelligent modeling and weight quantification module constructs a multiple linear regression model with the research results as the dependent variable and the impact factors as the independent variables. It iteratively fits the model using the least squares method and an AI adaptive mechanism, and calculates the weights based on the standardized transformation of the regression coefficients. The automatic sorting and visualization output module sorts the data in descending order based on the weight values, displays the weight distribution chart and factor ranking table on the web front end, and exports the analysis report.

[0010] Furthermore, the multi-source data preprocessing module performs the following steps: Step a1, Data Access and Intelligent Filling of Missing Values: Obtain the multi-dimensional raw dataset uploaded by the user, use the mean of the same column to fill for numerical variables, and use the mode to fill for categorical variables, and automatically fill in the missing data; Step a2, Outlier Removal Based on Interquartile Range: A robust statistical algorithm is used to calculate the interquartile range of each feature column. Identify and automatically remove items exceeding the limit. and the lower limit is Outlier anomalous samples; Step a3, Z-score standardization and normalization: The calculation formula is as follows: ; in, These are the raw values ​​of the impact factor. The mean of the feature column containing the influencing factor. The standard deviation of the feature column containing the influencing factor is used to map the influencing factors of different dimensions and orders of magnitude to a standard normal distribution space with a mean of 0 and a standard deviation of 1.

[0011] Furthermore, the intelligent modeling and weight quantization module performs the following steps: Step b1: Construction of a multiple linear regression model: using the target research outcome as the dependent variable. Using various influencing factors as independent variables , … Construct a global linear mapping model ; Step b2, AI Adaptive Regularized Fitting: An L2 regularization term is introduced into the loss function of the multiple linear regression model to construct the ridge regression loss function: ; in, For the actual value of the i-th data set, Let i be the predicted value of the i-th data set. Let j be the j-th regression coefficient in the ridge regression loss function. The system has a built-in AI adaptive optimization mechanism that dynamically traverses the regularization parameters through K-fold cross-validation. The space is used to determine the optimal penalty coefficient with the goal of minimizing the mean square error of the verification set. A dynamic balance is achieved between suppressing overfitting and preserving feature information, thus completing the iterative fitting of the model. Step b3, Intelligent Weight Quantization and Decoupling Calculation: Based on the partial regression coefficients output by the optimal model, and combined with the standard deviations of the independent and dependent variables, the standardized regression coefficients are calculated. ; in, The absolute value of the standardized regression coefficient. These are the partial regression coefficients output by the optimal model. Let the standard deviation of the independent variable be . The standard deviation of the dependent variable is the absolute value of the standardized regression coefficient. Weights of each influencing factor .

[0012] Furthermore, the automatic sorting and visualization output module performs the following steps: Step c1, Automatic Priority Sorting: Based on the calculated weight values The system automatically triggers a sorting algorithm to sort the influencing factors in descending order of their weight values, and outputs a list of influencing factors with clear priorities. Step c2, Visualization and Closed-Loop Report Generation: The web front-end renders a weight distribution bar chart and a factor ranking table, automatically generating a report containing model evaluation metrics. , The report analyzes the regularization parameters and the weights of each influencing factor, and exports the analysis report.

[0013] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of a multi-factor weighted AI analysis model and method based on linear regression.

[0014] The advantages of this invention compared to the prior art are: This invention maps influencing factors with different physical meanings and magnitudes to a dimensionless weight space through Z-score standardization and the conversion of standardized regression coefficients. This completely eliminates dimensional interference and enables heterogeneous factors such as drug dosage and extraction temperature to have absolute objective comparability for the first time. It achieves objective and accurate quantification and horizontal comparison of factor contribution, and truly realizes a paradigm upgrade from subjective qualitative assessment based on experience to objective quantitative assessment based on algorithms.

[0015] This invention combines L2 regularization with K-fold cross-validation, giving the model AI adaptive optimization capabilities. The dynamic adjustment of the penalty coefficient effectively reduces the coefficient shift of collinear variables, ensuring the robustness and authenticity of weight calculation, and adaptively preventing overfitting of the model on small samples or high-dimensional data. This significantly improves the model's generalization ability and fundamentally eliminates weight distortion caused by coefficient distortion.

[0016] This invention integrates outlier removal based on interquartile range, intelligent filling of mean and mode, standardization transformation, modeling and weight ranking into an end-to-end automated pipeline. The entire analysis process requires no human intervention, eliminating the possibility of researchers selectively manipulating the data to meet expected conclusions, and ensuring the absolute objectivity and complete reproducibility of the analysis results.

[0017] This invention is based on a Web cloud architecture, which encapsulates complex statistical tests, regularization iterations, matrix solving and visualization rendering at the underlying level. Users only need to upload data and click once to obtain a professional-grade visualization report containing model evaluation indicators, factor weights and priority rankings within seconds. This reduces the traditional analysis cycle of hours or even days to seconds, allowing people without a statistical background to easily obtain expert-level data insights. Attached Figure Description

[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application.

[0019] In the attached diagram: Figure 1 This is an overall flowchart of a multi-factor weighted AI analysis method based on linear regression according to the present invention. Detailed Implementation

[0020] The following detailed description of the embodiments is used to illustrate the principles of this application, but should not be used to limit the scope of this application. That is, the multi-factor weight AI analysis model and method based on linear regression in this application are not limited to the described embodiments.

[0021] The present invention will be further described below with reference to embodiments.

[0022] In specific embodiments, such as Figure 1 As shown, this invention uses a web page front-end as the interactive platform. After the user uploads a data file, the system automatically completes the entire analysis process, including data access, preprocessing, modeling, weight quantification, sorting, and visualization output. Data files support formats such as CSV and Excel, where the first row contains variable names, the last column contains the target variable, and the sample size is no less than 5 groups. The system automatically handles missing values. Analysis modes include fully automatic analysis and manual configuration. In fully automatic analysis mode, the system automatically selects the model based on the sample size: Ridge regression with L2 regularization is used to prevent overfitting when the sample size is small; Elastic Network is used to balance bias and variance when the sample size is medium; Gradient Boosting Tree is used when the sample size is large and the number of features is small; and Random Forest is used when the number of features is large. In manual configuration mode, the user can specify multiple linear regression, Ridge regression, Lasso regression, or Elastic Network. The system employs algorithms such as random forest or gradient boosting, and allows for settings for outlier handling, data standardization, and cross-validation folds. After standardizing the data, the system trains the model and evaluates its generalization ability using K-fold cross-validation or leave-one-out cross-validation. It calculates model evaluation metrics such as coefficient of determination, adjusted coefficient of determination, root mean square error, and mean absolute percentage error. Based on the model coefficients or feature importance, it calculates the standardized weights of each influencing factor, sorts them in descending order of weight, and renders a weight distribution map, factor ranking table, and correlation heatmap on the web front-end. The system generates an analysis report containing model evaluation metrics, factor weights, proportions, and rankings, and supports export.

[0023] Example 1: Multifactor Weight Analysis of the Hypoglycemic Efficacy of Traditional Chinese Medicine Extracts In specific embodiments, such as Figure 1 As shown, the analytical model of this invention integrates data cleaning, modeling, weight quantification, sorting, and visualization into an automated process. In a study on the hypoglycemic efficacy of a certain traditional Chinese medicine extract, researchers collected 200 sets of experimental data, including five independent variables that may affect the hypoglycemic effect: X1 dosage, X2 extraction temperature, X3 extraction time, X4 particle size of the medicinal material, and X5 solvent concentration. The extraction temperature is expressed in degrees Celsius, the particle size of the medicinal material is expressed in mesh, the solvent concentration is expressed as a percentage, and the percentage decrease in blood glucose is used as the dependent variable Y.

[0024] Step 1: Data Access and Preprocessing. After uploading the data to the online platform of this invention, the system automatically performs preprocessing. In this embodiment, X5 was found to be missing in 3 sets of data. The system automatically filled the missing data with 75% of the mean of the X5 column to ensure sample integrity. Subsequently, a robust statistical algorithm was used to calculate the interquartile range of each feature column, identify and automatically remove outlier samples that exceed the upper and lower limits to prevent noise from interfering with model fitting. In this embodiment, 2 sets of data were found to have abnormally high values ​​and were automatically removed, leaving 198 valid samples. Due to the large difference in dimensions between the drug dosage and the extraction temperature, the system standardized the data according to the Z-score formula. Each influencing factor is mapped to a standard normal distribution space with a mean of 0 and a standard deviation of 1, thus eliminating the interference of dimensional differences on subsequent regression coefficients.

[0025] Step 2: Construct a multiple linear regression AI model and quantify weights. The system uses the percentage decrease in blood glucose as the dependent variable and five pre-processed factors as independent variables to construct a multiple linear regression model. The model form is as follows: To overcome coefficient shifts caused by multicollinearity, the system introduces an L2 regularization term to construct a ridge regression loss function. By dynamically traversing the regularization parameter space through 5-fold cross-validation, and aiming to minimize the mean square error of the validation set, the optimal penalty coefficient is automatically locked. This effectively suppresses overfitting caused by multicollinearity, achieving a dynamic balance between suppressing overfitting and preserving feature information, completing iterative model fitting, and determining the model fitting coefficient. The root mean square error reached 0.89. The coefficient of variation was 4.52%, indicating a good model fit. Subsequently, the partial regression coefficients after fitting were extracted, and the standard deviations of the independent and dependent variables were combined and substituted into the standardized regression coefficient formula. Calculate the standardized regression coefficients of each factor and take their absolute values ​​as the weights of the true influence, where the dosage corresponds to... Take its absolute value as the weight. Solvent concentration corresponding to Take its absolute value as the weight. Extraction temperature corresponding to Take its absolute value as the weight. Extraction time corresponding to Take its absolute value as the weight. The corresponding particle size of medicinal materials Take its absolute value as the weight. This will enable the objective quantification of the independent physical contributions of all influencing factors.

[0026] Step 3: Sorting and Result Output. The system automatically sorts the calculated weight values ​​in descending order and outputs an impact factor list: Dosage (0.58) > Solvent Concentration (0.32) > Extraction Temperature (0.21) > Extraction Time (0.09) > Herb Particle Size (0.04). The front-end page renders a weight distribution bar chart in real time and automatically generates a PDF analysis report containing model evaluation indicators, weights of each factor, and ranking.

[0027] Conclusion Analysis: The model of this invention clearly quantifies that the dosage is the core factor determining the hypoglycemic effect, with a weight of 0.58, while the particle size of the medicinal materials has a negligible impact, with a weight of 0.04. This result is objective and reproducible, avoiding the ambiguity of subjective human judgment, and directly guiding the direction of subsequent experimental optimization, namely, prioritizing the adjustment of the dosage and solvent concentration.

[0028] Example 2: Analysis of Factors Affecting the Stability of Antibacterial Drugs In specific embodiments, such as Figure 1 As shown, this embodiment uses the stability study of a certain antibacterial drug formulation as an example to verify the universality and automation advantages of this model in multivariate studies.

[0029] Step 1: Data Access. The system imports 150 sets of drug stability data. The dependent variable is the drug's shelf life, and the independent variables include storage temperature, ambient humidity, packaging seal, and excipient addition. The drug's shelf life is measured in months, the storage temperature in degrees Celsius, the ambient humidity in percentage of relative humidity, and the packaging seal in a score from 0 to 10.

[0030] Step 2: Model Training and Weight Quantization. The system constructs a multiple linear regression model with the drug's effective period as the dependent variable and four preprocessed factors as independent variables. The model form is as follows: To overcome coefficient shifts caused by multicollinearity, the system introduces an L2 regularization term to construct a ridge regression loss function. The AI ​​adaptive mechanism identifies the collinearity between humidity and sealing, automatically increasing the regularization parameter for penalized smoothing, achieving a dynamic balance between suppressing overfitting and preserving feature information, and completing the model iterative fitting. Subsequently, the system extracts the fitted partial regression coefficients, combines the standard deviations of the independent and dependent variables, and substitutes them into the standardized regression coefficient formula. The standardized regression coefficients of each factor are calculated, and their absolute values ​​are taken as the weights of the true impact. The final standardized weights output by the model are as follows: (Storage temperature) = 0.65 (Ambient humidity) = 0.42 (Packaging airtightness) = 0.38 (Amount of excipients added) = 0.12.

[0031] Step 3: Sorting and Visualization. The system automatically outputs the sorting results: storage temperature > ambient humidity > packaging sealing > excipient addition amount. At the same time, a weighted heat map is generated to intuitively show the absolute dominance of temperature on drug stability.

[0032] Comparative verification shows that in traditional analysis, researchers rely on experience to believe that packaging airtightness is the primary factor in preventing moisture and deterioration. However, this model reveals through algorithmic quantification that, in the current dataset, the weight of storage temperature (0.65) far exceeds the weight of packaging airtightness (0.38), correcting the bias of human experience-based judgment. Furthermore, the entire analysis process, from data upload to report export, takes only 12 seconds, while using traditional SPSS software to complete the same analysis requires at least 2 hours of data cleaning and operation. This fully demonstrates the advantages of this invention: lightweight, automated, and objective quantification.

[0033] It should be noted that the combination of the technical features in this case is not limited to the combination methods described in the claims of this case or the combination methods described in the specific embodiments. All technical features described in this case can be freely combined or combined in any way, unless they contradict each other.

[0034] It should also be noted that the embodiments listed above are merely specific embodiments of the present invention. Obviously, the present invention is not limited to the above embodiments, and similar changes or modifications made thereto are those that can be directly derived or easily conceived by those skilled in the art from the content disclosed in the present invention, and should all fall within the protection scope of the present invention.

Claims

1. A multi-factor weighted AI analysis method based on linear regression, characterized in that, Includes the following steps: Step S1, Multi-source data integration and preprocessing: Obtain multi-dimensional raw data uploaded by users, automatically perform missing value filling, outlier removal and data normalization processing, eliminate differences in units, and standardize the input of multi-factor data; Step S2: Construct a multiple linear regression AI model: Using the target research outcome as the dependent variable and various influencing factors as independent variables, construct a multiple linear regression model. Iterate and fit the model using the least squares method combined with the AI ​​adaptive iterative optimization mechanism to establish the correlation between factors and results. Step S3, Intelligent Weight Automatic Calculation: Based on the model regression coefficients, standardization transformation is performed to quantify the true impact weight of each factor, and the contribution of all influencing factors is objectively quantified; Step S4: Automatic Priority Ranking of Influence Factors: Based on the calculated weight values, the influencing factors are automatically sorted in descending order from largest to smallest, distinguishing between core influencing factors and weak influencing factors. Step S5, Result Visualization and Data Output: Display the calculation results through charts, weight lists, and analysis reports, and export the data and generate reports.

2. The multi-factor weighted AI analysis method based on linear regression according to claim 1, characterized in that, In step S1, the data normalization process uses Min-Max standardization or Z-score standardization to map influence factors of different dimensions and orders of magnitude to a unified interval.

3. The multi-factor weighted AI analysis method based on linear regression according to claim 1, characterized in that, In step S2, the AI ​​adaptive iterative optimization mechanism is as follows: during the least squares fitting process, the model parameters are dynamically adjusted, a regularization term is introduced to prevent overfitting, and the optimal feature combination is adaptively selected based on the characteristics of the dataset.

4. The multi-factor weighted AI analysis method based on linear regression according to claim 1, characterized in that, In step S3, the standardization transformation specifically involves multiplying the partial regression coefficients obtained from the multiple linear regression model by the ratio of the standard deviation of the corresponding independent variable to the standard deviation of the dependent variable, thereby converting them into standardized regression coefficients. The absolute value of the standardized regression coefficients is then used as the true influence weight of each factor.

5. A multi-factor weighted AI analysis model based on linear regression, characterized in that, It includes a multi-source data preprocessing module, an intelligent modeling and weight quantization module, and an automatic sorting and visualization output module; The multi-source data preprocessing module acquires multi-dimensional raw data in batches and automatically completes missing value filling, outlier removal, and data normalization. The intelligent modeling and weight quantification module constructs a multiple linear regression model with the research results as the dependent variable and the impact factors as the independent variables. It iteratively fits the model using the least squares method and an AI adaptive mechanism, and calculates the weights based on the standardized transformation of the regression coefficients. The automatic sorting and visualization output module sorts the data in descending order based on the weight values, displays the weight distribution chart and factor ranking table on the web front end, and exports the analysis report.

6. The multi-factor weighted AI analysis model based on linear regression according to claim 5, characterized in that, The multi-source data preprocessing module performs the following steps: Step a1, Data Access and Intelligent Filling of Missing Values: Obtain the multi-dimensional raw dataset uploaded by the user, use the mean of the same column to fill for numerical variables, and use the mode to fill for categorical variables, and automatically fill in the missing data; Step a2, Outlier Removal Based on Interquartile Range: A robust statistical algorithm is used to calculate the interquartile range of each feature column. Identify and automatically remove items exceeding the limit. and the lower limit is Outlier anomalous samples; Step a3, Z-score standardization and normalization: The calculation formula is as follows: ； in, These are the raw values ​​of the impact factor. The mean of the feature column containing the influencing factor. The standard deviation of the feature column containing the influencing factor is used to map the influencing factors of different dimensions and orders of magnitude to a standard normal distribution space with a mean of 0 and a standard deviation of 1.

7. The multi-factor weighted AI analysis model based on linear regression according to claim 5, characterized in that, The intelligent modeling and weight quantization module performs the following steps: Step b1: Construction of a multiple linear regression model: using the target research outcome as the dependent variable. Using various influencing factors as independent variables , … Construct a global linear mapping model ; Step b2, AI Adaptive Regularized Fitting: An L2 regularization term is introduced into the loss function of the multiple linear regression model to construct the ridge regression loss function: ； in, For the actual value of the i-th data set, Let i be the predicted value of the i-th data set. Let j be the j-th regression coefficient in the ridge regression loss function. The system has a built-in AI adaptive optimization mechanism that dynamically traverses the regularization parameters through K-fold cross-validation. The space is used to determine the optimal penalty coefficient with the goal of minimizing the mean square error of the verification set. A dynamic balance is achieved between suppressing overfitting and preserving feature information, thus completing the iterative fitting of the model. Step b3, Intelligent Quantization and Decoupling Calculation of Weights: Based on the partial regression coefficients output by the optimal model, and combined with the standard deviations of the independent and dependent variables, the standardized regression coefficients are calculated. ； in, The absolute value of the standardized regression coefficient. These are the partial regression coefficients output by the optimal model. Let the standard deviation of the independent variable be . The standard deviation of the dependent variable is the absolute value of the standardized regression coefficient. Weights of each influencing factor .

8. The multi-factor weighted AI analysis model based on linear regression according to claim 5, characterized in that, The automatic sorting and visualization output module performs the following steps: Step c1, Automatic Priority Sorting: Based on the calculated weight values The system automatically triggers a sorting algorithm to sort the influencing factors in descending order of their weight values, and outputs a list of influencing factors with clear priorities. Step c2, Visualization and Closed-Loop Report Generation: The web front-end renders a weight distribution bar chart and a factor ranking table, automatically generating a report containing model evaluation metrics. , The report analyzes the regularization parameters and the weights of each influencing factor, and exports the analysis report.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it performs each step of the multi-factor weight AI analysis method based on linear regression as described in any one of claims 1 to 4.

Images