Statistical analysis method for digital transformation of automobile manufacturing
By analyzing the financial and annual report data of listed companies in the automobile manufacturing industry, building a digital transformation index, and using the support vector machine regression model, the problem of measuring market competitiveness in the era of big data is solved, and the market competitiveness has been effectively improved and prediction.
Patent Information
- Application Number
- CN202411104491.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-11-19
AI Technical Summary
In the era of big data and artificial intelligence, how can listed automobile manufacturing companies effectively measure and enhance their market competitiveness, especially in the process of digital transformation, it is difficult for existing technologies to fully reflect the multi-dimensional transformation and competitiveness of enterprises.
By collecting and analyzing the financial data and annual report texts of listed companies in the automobile manufacturing industry, building a digital transformation index of enterprises, combining the support vector machine regression model, we explore the impact of the degree of digital transformation on market competitiveness, and analyze and control variables through Pearson's correlation to establish a relationship between market competitiveness and the degree of digital transformation.
The nonlinear regression problem was effectively dealt with, and market competitiveness was improved. By simulating different digital transformation levels, market competitiveness was predicted, the significant positive effect of digital transformation on competitiveness was demonstrated, and data was provided to support investment decisions and market analysis.
Smart Images

Figure CN118982148A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a statistical analysis method, and in particular to a statistical analysis method for digital transformation of automobile manufacturing. Background Art
[0002] In the era of big data, listed companies in the automotive manufacturing industry are facing unprecedented opportunities and challenges. With the rapid development of information technology, data has become one of the core assets of enterprises, which has a profound impact on the decision-making efficiency, operation mode and market competitiveness of enterprises.
[0003] In recent years, with the development of high-tech technologies such as big data, artificial intelligence, and cloud computing, the development of the automotive manufacturing industry has been deeply empowered. These technologies make real-time data analysis, intelligent decision support, remote monitoring, and predictive maintenance possible, greatly improving production efficiency, reducing operating costs, and promoting product and service innovation.
[0004] The current consumer demand for automobiles is becoming increasingly personalized and diversified, which has forced automobile companies to enhance market response speed through digital means, achieve product customization and service differentiation. Digital transformation can help companies better collect and analyze customer data, understand market demand, and quickly adjust product design and marketing strategies. Therefore, this paper focuses on the relationship between the digital transformation of listed companies in the automobile manufacturing industry and their market competitiveness in the context of the era of big data and artificial intelligence. This reflects the impact of high-tech technologies such as big data and artificial intelligence on the comprehensive competitiveness of listed companies in the automobile manufacturing industry in the context of the era of big data and artificial intelligence. Summary of the invention
[0005] In view of the shortcomings of the prior art, the present invention provides a statistical analysis method for the digital transformation of automobile manufacturing. We collected basic information such as the year of listing, monetary funds, total current assets, etc. of the company, and collected relevant financial data such as total current liabilities and total fixed liabilities. These data were quantified into variables such as enterprise stress resistance (EPR), goodwill (GA), cash flow level (CF), financial leverage (FZ), knowledge economy level (LOKE), and enterprise scale (ES) according to different weights and indicators. The above variables are used as explanatory variables, and product market competitiveness (PMC) is the explained variable. The data are then processed to explore the impact of the degree of "digital transformation" of the enterprise on its comprehensive market competitiveness.
[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A statistical analysis method for the digital transformation of automobile manufacturing, which selects the listed company data from 2010 to 2022 as a research sample and performs the following preliminary screening on the data:
[0007] 1 Eliminate samples with incomplete data or data with significant errors;
[0008] 2 Eliminate samples where the industry code does not exist;
[0009] 3 Eliminate ST and *ST enterprise samples.
[0010] In the era of big data, the digital transformation of companies in the automotive manufacturing industry has had a more obvious impact on the industry. We screened their data and selected sample data from 35 listed companies in the automotive manufacturing industry for research and analysis.
[0011] Preferably, the core explanatory variable of this paper is the digital transformation (DT) of enterprises. The digital transformation of enterprises is a multi-dimensional and complex process, covering many aspects such as enterprise management and production processes, so it is difficult to fully measure it with only a single financial indicator.
[0012] Drawing on the research methods of existing scholars, this paper deeply analyzes the frequency of occurrence of key words related to enterprise digital transformation, and then uses these word frequency data to construct a comprehensive enterprise digital transformation index. This index can more comprehensively and accurately reflect the actual situation and progress of enterprises in the process of digital transformation.
[0013] The specific steps are as follows:
[0014] (1) Using Python programming technology, we systematically collected the annual financial reports of 35 listed companies in the automotive manufacturing industry. Through this detailed data collection process, we focused on the "Management Discussion and Analysis" section in the annual report, which is usually rich in important information about the company's development direction, strategy adjustments, and market prospects, thus providing us with valuable clues for in-depth analysis of the company's growth trajectory and future planning. Through this method, we can more accurately grasp the development trends of these listed companies in the automotive manufacturing industry and provide strong data support for investment decisions and market analysis.
[0015] (2) This paper conducts an in-depth study on the frequency of appearance of five key areas - artificial intelligence, blockchain, cloud computing, big data and digital technology applications - in corporate annual reports. We accurately extracted keywords for each of these areas and systematically measured them through detailed indicator divisions (as shown in Table 1). In order to comprehensively evaluate the degree of digital transformation (DT) of an enterprise, we summarized the word frequencies of the enterprises in these five categories in that year and constructed a dictionary of corporate digital terms. Given that these word frequency data show a "right-skewed" characteristic, in order to make them more in line with the needs of statistical analysis, we further performed logarithmic processing to more accurately and scientifically reflect the degree of digital transformation of the enterprise. The specific calculation formula is as follows:
[0016] DT=ln(AI+BC+CC+BD+AODT+1) (1)
[0017] Table 1 Detailed indicators of key areas of digital transformation
[0018]
[0019]
[0020] (3) We used Python technology to extract in-depth data from the annual reports of listed companies. We then searched and matched them accurately based on the keywords shown in Table 1. By using the word frequency counting method, we effectively counted the frequency of vocabulary in each key technical direction and classified these word frequencies. Finally, we added up the word frequencies of each category to build a set of indicators that comprehensively reflect the status of digital transformation of enterprises.
[0021] Preferably, in order to improve the accuracy of the study, this paper adds a series of control variables based on the data.
[0022] Enterprise stress resilience (EPR) is an indicator that measures a company's ability to maintain robust operations in the face of challenges and adversities. When evaluating this indicator, we use a simple and intuitive method: calculate the difference between the company's time to market (MT) and the current year (CT). The larger this difference is, the more market ups and downs the company has experienced, the more experience and resources it has accumulated, and thus shows stronger stress resilience. In short, the longer the listing time, the greater the company's stress resilience tends to be, which reflects its ability to maintain robust operations in the face of various challenges. The formula is as follows:
[0023] EPR=CT-MT (2)
[0024] Goodwill amount (GA) is an important indicator used to measure the potential economic value of listed companies. In the calculation process, we selected the net goodwill (NG) of the company as the benchmark data and expressed this amount by taking its natural logarithm. This processing method can more intuitively and accurately reflect the potential economic value of the company. At the same time, considering the possible negative situation of net goodwill, we adopted the strategy of taking the natural logarithm of the absolute value to ensure the rationality and accuracy of the calculation results. If the net goodwill is a negative number, we will add a minus sign before the final result to retain its original negative meaning. Through this processing method, the high or low GA value can directly reflect the strength of the company's potential economic value. The formula is as follows:
[0025]
[0026] Cash flow level (CF) is a key indicator for assessing the health and operating efficiency of a company. In order to accurately measure this level, we select the company's current year monetary amount (MA) as the core data. By taking the natural logarithm of this amount, we can get a more intuitive and standardized cash flow level indicator. This processing method can not only reflect the absolute value of the company's cash flow, but also eliminate the data differences between companies of different sizes, making the cash flow levels between different companies more comparable. Therefore, the level of cash flow (CF) can directly reflect the efficiency and robustness of the company's capital operation. The formula is as follows:
[0027] CF=lnMA (4)
[0028] Financial leverage (FZ) is a key indicator that measures the use of debt financing by a company to support its operating activities. To calculate this indicator, we use a specific method. First, we add the company's current liabilities (CI) and long-term liabilities (LTL) to get the amount of total liabilities of the company. Next, we divide this total debt amount by the company's total assets (TEA) to get the specific value of financial leverage. This process is essentially evaluating the ratio of a company's debt to the size of its assets, reflecting the extent to which the company uses debt capital. The formula is as follows:
[0029]
[0030] The level of knowledge-based economy (LOKE) is an important indicator to measure the ability of enterprises to create value by using knowledge resources in the knowledge economy era. In the calculation of this indicator, we selected the intangible assets (IA) of the enterprise as the core data. Intangible assets, such as patents, trademarks, non-patented technologies, copyrights and land use rights, are important components of corporate assets. They can reflect the investment and achievements of enterprises in technological innovation, brand building, intellectual property protection, etc., and thus reflect the level of knowledge-based economy of enterprises. In order to more intuitively display the level of knowledge-based economy of enterprises, we have taken the natural logarithm of intangible assets. This processing method can not only retain the continuity of the intangible asset values, but also eliminate the data differences between enterprises of different sizes, making the levels of knowledge-based economy between different enterprises more comparable. The formula is as follows:
[0031] LOKE=ln(IA) (6)
[0032] Enterprise size (ES) is an important dimension for evaluating the strength and operating capacity of an enterprise. In order to accurately measure this indicator, we selected total enterprise assets (TEA) as the benchmark data. Total assets are the sum of all assets of an enterprise, which can fully reflect the resource reserves and operating scale of the enterprise. However, due to the large differences in the total asset scale of different enterprises, direct comparison is difficult to reflect the relative size of enterprises. Therefore, we use the natural logarithm method to process the total assets of the enterprise, so as to obtain a more intuitive and comparable enterprise size indicator. The formula is as follows:
[0033] ES=ln(TEA) (7)
[0034] Preferably, product market competitiveness (PMC) is a key indicator that comprehensively reflects the company's competitive situation. This indicator can reveal the company's advantages and disadvantages in pricing strategy, investment layout and other competitive means by capturing the changes in the company's sales (SV) relative to the entire industry. In order to quantify this competitiveness more accurately, we use the natural logarithm change of the company's sales (SV) as a measure. This method can not only reflect the increase or decrease in sales, but also eliminate the heteroscedasticity in the data, thereby more accurately reflecting the company's true competitiveness in the product market. The formula is as follows:
[0035] PMC=ln(SV) (8)
[0036] Preferably, during the data processing, we adopted specific strategies to ensure the completeness and accuracy of the data in response to the missing value problem in the collected data. Specifically, for variables with missing values, especially those related to the performance of car companies in different years, we chose to use the average value of the variable in all years for the car company to fill it. This method is based on the assumption that the performance of car companies on the same variable has a certain degree of stability or trend over time, so using the average value can provide a relatively reasonable estimate.
[0037] In addition to missing value processing, we also formatted the data. Formatting is to ensure that the data is in a uniform format and easy to analyze and use. This may include converting the data into a uniform date format or currency format, or renaming or reordering the data to better suit our analysis needs.
[0038] Preferably, we extract the basic statistical characteristics of the main variables, as shown in the following table:
[0039] Table 2 Basic statistical characteristics of various variables
[0040]
[0041] Preferably, Pearson correlation analysis is a statistical method for measuring the strength and direction of the linear relationship between two continuous variables. It measures the correlation between the variables based on the covariance between them and standardizes them to obtain the correlation coefficient. The value range of the Pearson correlation coefficient is between -1 and 1.
[0042] If there are two sets of population data X: {X 1 , X 2 , ..., X n} and Y: {Y 1 , Y 2 , ..., Y n}, then calculate their overall covariance Cov(X, Y), the formula is as follows:
[0043]
[0044] Where E(Y) and E(X) are population means.
[0045] At the same time, calculate the standard deviation σ of X and Y X and σ Y , from the above formula we can get:
[0046]
[0047] Through the above formula, we can get the Pearson correlation coefficient expression ρ XY , as shown below,
[0048]
[0049] Usually, the correlation strength of variables is judged by the value range in Table 3:
[0050] Table 3 Correlation strength indicator table between variables
[0051]
[0052] Preferably, the support vector machine (SVM) solves the classification or regression task by maximizing the margin idea. In this paper, we apply the support vector machine regression (SVR) to solve the model.
[0053] The sample points within the tolerance region are called support vectors. Assuming a data set of size n is given, the linear regression decision surface function of SVR is as follows:
[0054] f(x i )=w T x i +b (12)
[0055] According to maximizing the hard margin of the tolerance region and minimizing the total loss, the learning expression is as follows:
[0056]
[0057] Among them, C>0 is the regularization penalty coefficient (hyperparameter) used to control the tolerance level, which is the loss function, defined as follows:
[0058]
[0059] Table 4 Statistics of common kernel functions of SVR
[0060]
[0061]
[0062] Preferably, in this study, the data collected from the automobile industry has the characteristics of approximate normal distribution and equal variance after processing. Therefore, we choose to use the Pearson correlation analysis model to explore whether there is a significant linear correlation between the market competitiveness of automobile companies and the degree of digital transformation (DT) of enterprises, and also examine the linear correlation between other explanatory variables.
[0063] In this study, we know through Pearson correlation analysis that the linear correlation between the core variable DT and other variables is weak, so we choose the support vector machine regression model for modeling analysis, and select the "Radial Basis Function" (RBF) as the kernel function of the support vector to build the model and determine its decision boundary. In order to optimize the performance of the model, we pay special attention to the two key parameters of the RBF kernel function-gamma and C, and make fine adjustments. At the same time, the PMC values of different companies are used as the target values of the model, and the remaining variables are trained as feature values. By adjusting the two parameters of gamma and C, we find an optimal SVM model that can show the highest accuracy on the test data.
[0064] We took an experimental approach to explore the relationship between enterprise digital transformation and market competitiveness in the context of big data. Specifically, we used a trained model to simulate different levels of digital transformation by artificially adjusting the metric of enterprise digital transformation (DT).
[0065] Next, we added 1 to the DT value and input these modified DT values into the model to predict the market competitiveness of the enterprise under this simulated transformation level. In this way, we can compare the predicted values output by the model with the competitiveness of the enterprise in the actual market.
[0066] The present invention discloses a statistical analysis method for the digital transformation of automobile manufacturing, which has the following beneficial effects:
[0067] 1. This statistical analysis method for the digital transformation of automobile manufacturing uses a support vector machine regression model to effectively handle nonlinear regression problems. SVR performs well in complex regression tasks. Compared with traditional regression methods, SVR has better robustness to outliers and noise. At the same time, the SVR algorithm is relatively simple, easy to implement and understand. At the same time, SVR can effectively process high-dimensional data of automobile companies because it is based on the principle of structural risk minimization and maps the input space to a high-dimensional feature space through a kernel function, making the data in the high-dimensional space linearly separable or easier to find a regression function.
[0068] 2. This statistical analysis method for the digital transformation of automobile manufacturing. In the era of big data, the digital transformation of enterprises is a powerful lever that can significantly improve market competitiveness, but its effect is affected by many factors such as the matching degree of transformation strategy, execution efficiency and external environment. Therefore, enterprises should carefully plan their digital transformation path to ensure that the transformation measures are closely aligned with the strategic goals of the enterprise to maximize the positive effects of the transformation and avoid potential negative effects. At the same time, a continuous monitoring and evaluation mechanism is essential for timely adjustment of transformation strategies and optimization of resource allocation to ensure that digital transformation can continuously and effectively drive the improvement of enterprise market competitiveness.
[0069] 3. This statistical analysis method for the digital transformation of automobile manufacturing, this study systematically explored the relationship between enterprise digital transformation (DT) and market competitiveness (PMC) under the background of big data through an experimental method. First, using Pearson correlation analysis, it was found that DT (digital transformation degree) and PMC (market competitiveness) have a not-so-significant positive linear correlation. Then, using the support vector machine model, the pre-trained model was used to simulate different levels of digital transformation and observe its impact on predicting market competitiveness. The analysis results show that the deepening of digital transformation has a significant positive effect on the market competitiveness of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS
[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0071] Figure 1 It is a hard interval regression analysis diagram of the present invention;
[0072] Figure 2 It is the soft interval regression analysis diagram of the present invention;
[0073] Figure 3 This is the data distribution diagram of the automobile industry of the present invention;
[0074] Figure 4 The Pearson coefficient heat map of the automobile enterprise related data of the present invention;
[0075] Figure 5 is the changing trend of the gamma parameter of the model of the present invention;
[0076] Figure 6 is the variation trend of the model parameter C of the present invention;
[0077] Figure 7 This is a PMC change trend analysis diagram based on DT changes in the present invention;
[0078] Figure 8 This is a trend analysis diagram of the PMC phase difference value change based on DT change in the present invention. DETAILED DESCRIPTION
[0079] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0080] The embodiment of the present invention discloses a statistical analysis method for digital transformation of automobile manufacturing;
[0081] Using Python programming technology, we systematically collected the annual financial reports of 35 listed companies in the automobile manufacturing industry from 2010 to 2022 as research samples, and revised and classified the data.
[0082] By summarizing the sample areas and conducting in-depth analysis on the frequency of occurrence of key words related to enterprise digital transformation, we use these word frequency data to construct a comprehensive enterprise digital transformation index, which provides valuable clues for in-depth analysis of the company's growth trajectory and future plans, and more accurately grasps the development trends of these listed companies in the automotive manufacturing industry, providing strong data support for investment decisions and market analysis.
[0083] We extracted keywords precisely for each characteristic field and systematically measured them through detailed indicator division. In order to comprehensively evaluate the degree of digital transformation (DT) of enterprises, we summarized the word frequencies of enterprises in these five categories in that year and constructed a dictionary of enterprise digital terms. Given that these word frequency data showed a "right-biased" characteristic, in order to make them more in line with the needs of statistical analysis, we further performed logarithmic processing, thereby more accurately and scientifically reflecting the degree of digital transformation of enterprises.
[0084] We make a preliminary interpretation of the statistical data in Table 2 and can draw the following observations:
[0085] When examining the degree of digital transformation of enterprises, the variable DT shows significant differences. Its maximum value is 6.64, while the minimum value is 0, and the median value of 2.89 is between the two. This data distribution reveals that there are large differences in digital transformation among different enterprises. This shows that some enterprises may have achieved a high level of digitalization, while others may still be in the initial stage.
[0086] In addition, we also noticed that a series of control variables such as EPR and GA also had significant differences in their maximum and minimum values. This observation shows that different companies also show significant differences in internal management and operational efficiency. These differences may be due to various factors such as the scale of the company, industry characteristics, and management strategy.
[0087] Furthermore, the maximum value of the variable PMC is 24.87, the minimum value is 17.00, and the mean value is 20.80. This data distribution reflects that there are certain differences in the market competitiveness of enterprises. A higher PMC value may mean that the enterprise has a stronger competitive advantage in the market, while a lower PMC value may indicate that the enterprise is in a relatively disadvantaged position in the market competition.
[0088] The following conclusions are drawn from the understanding of the application of the Pearson correlation coefficient:
[0089] (1) When the correlation coefficient is 0, the X variable and the Y variable are unrelated;
[0090] (2) When the value of X and the value of Y increase or decrease at the same time, the two variables are positively correlated, and the correlation coefficient is between 0 and 1;
[0091] (3) When the value of X increases and the value of Y decreases, or when the value of X decreases and the value of Y increases, the two variables are negatively correlated, and the correlation coefficient is between -1 and 0.
[0092] Support vector regression (SVR) solves the regression model by introducing a tolerance region and maximizing the interval of the tolerance region while minimizing the total loss. It is divided into hard margin regression and soft margin regression. The interval of the maximum tolerance region is controlled by artificially setting the ε deviation, such as Figure 1 As shown in the figure, the sample points within the tolerance area are not included in the loss, while the sample points outside the tolerance area need to be included in the loss.
[0093] Similarly, for some abnormal points and noise points, it is difficult for the support vector hard margin regression model to determine the appropriate ε to ensure that most of the data are within the tolerance area, which seriously affects the prediction effect of the regression model. For this, soft margin regression is introduced, such as Figure 2 As shown, a slack variable is introduced into each sample to expand the tolerance region.
[0094] Then (13) can be rewritten as:
[0095]
[0096] Where C>0 is used as a regularization penalty coefficient (hyperparameter) to coordinate the relationship between w and (ξ', ξ).
[0097] The kernel method solves the problem of linearly inseparable input data by mapping the linearly inseparable data in the input space to a high-dimensional feature space, so that the input data can be linearly separable in the high-dimensional features, where the mapping function is defined as φ(X).
[0098] According to the optimization algorithm for solving convex quadratic programming with support vectors, it can be seen that there is X in the solution term i T X j ;
[0099] The input data is mapped to a high-dimensional space through a mapping function to solve the linearly inseparable task, that is,
[0100]
[0101] In order to avoid the problem of computational difficulties caused by the explosive increase in dimensions due to the mapping function, the kernel method defines the relevant kernel function, namely
[0102] κ(X i ,X j )=φ(X i ) T kφ(X j ) (17)
[0103] Therefore X i and X j The inner product mapping in the feature space is obtained by passing the kernel function κ(X i , X j) calculation is completed, avoiding the inner product of the high-dimensional feature space. For the common kernel functions in support vector machine regression (SVR), we generally have four types, namely Linear, poly, sigmoid and rbf. As shown in Table 4, Linear and poly generally solve more linear problems, while sigmoid and rbf generally solve more nonlinear problems.
[0104] Pearson correlation analysis is calculated by standardizing the covariance of two variables and then dividing it by the product of their respective standard deviations to determine the degree of linear correlation between the two variables. We bring all variables related to the development of automobile companies and product market competitiveness (PMC) into the Pearson correlation analysis model for solution. The solution results are as follows: Figure 4 shown.
[0105] From the heat map of the solution results, we can see that when the correlation between DT and a series of control variables (such as EPR, GA, CF, LOKE, ES) is shown, these correlation coefficients are in the range of 0.2 to 0.4, indicating that the linear relationship between DT and these variables is relatively weak. In addition, the correlation between DT and FZ is almost 0, which further emphasizes that there is almost no linear association between DT and FZ. The coefficient between PMC and DT is only 0.38, which shows that there is only a weak linear correlation between PMC and DT. At the same time, it shows that due to the weak linear relationship between DT and the control variables and PMC, simply applying a linear model may not be able to fully capture the complex relationship between these variables. Therefore, the linear model may not be the most ideal way to explain the relationship between these variables, and more complex models or analysis methods may need to be considered to more accurately describe and predict the relationship between these variables.
[0106] We tuned the gamma parameter and the results are shown in the attached figure. Figure 5 shown.
[0107] As attached Figure 5 As shown, we can find that when the gamma parameter reaches 0.0339, the accuracy of the model has reached 82.53%.
[0108] We tuned the C parameter based on gamma, and the results are shown in the attached figure. Figure 6 shown.
[0109] As attached Figure 6 As shown, when the parameter C is 23.2676, the accuracy of the model is the highest. At the same time, when the size of C is increased, the accuracy of the model remains basically unchanged, and the maximum accuracy is 87.96%.
[0110] After a series of parameter tuning and model training, the model we have built has achieved an accuracy of 87.96% on the test set, which is quite close to or meets our expected requirements for model performance. This indicates that our model has a good performance in prediction tasks and proves that our efforts in data processing, feature selection, and model selection have achieved significant results.
[0111] pass Figure 7 We observed that after adjusting the DT value, the market competitiveness (PMC) of different companies changed significantly. In order to capture this change more intuitively, we adopted a new analysis method. Specifically, we calculated the difference between the PMC value after the DT value was modified and the original PMC value, and visualized these differences. In this way, we can clearly see the specific impact of the adjustment of the DT value on the change of the PMC value of different companies. This result is shown in the attached figure. Figure 8 A visual display was given.
[0112] Combined with Figure 8 , we conduct statistical analysis on the data of PMC phase difference values, as shown in the following table:
[0113] Table 5 Basic statistical characteristics of PMC phase difference values
[0114]
[0115] As shown in the table above, when we artificially added value to the DT value (i.e., the degree of digital transformation) to simulate different degrees of digital transformation of enterprises, we found that these changes had a significant difference in the market competitiveness of enterprises. Specifically, the maximum value of the market competitiveness index difference reached 2.86, which shows that in some cases, the deepening of digital transformation can greatly enhance the market position and competitiveness of enterprises. At the same time, we also noticed that there was a minimum value of -1.18, which may be due to the fact that digital transformation failed to bring the expected market benefits in certain specific situations (such as improper transformation strategy, unequal resource allocation, etc.). However, from the overall trend, the benefits brought by enterprises' increased efforts in digital transformation are positive. This means that with the improvement of the degree of digital transformation, the market competitiveness of enterprises has generally been improved, which is obvious both from the average level and the overall distribution. This shows that digital transformation is of great significance for enterprises to maintain their competitiveness in the current big data context, and it is worthwhile for enterprises to invest more resources and energy to promote it.
Claims
1. A statistical analysis method for digital transformation of automobile manufacturing, characterized in that: The following steps are involved: S1. Obtain relevant sample data and perform data screening; S2. Analyze the core explanatory variables and construct a comprehensive enterprise digital transformation index; S3, introduce the Pearson correlation model, a statistical method for analyzing the strength and direction of linear relationships between continuous variables; S4. Solve the model through the support vector machine regression model, select and optimize the model parameters, and verify the accuracy of the analysis model.
2. The statistical analysis method for digital transformation of automobile manufacturing according to claim 1, characterized in that: Considering the correctness and authenticity of the final experimental results and reducing the influence of accidental factors, the data collected in step S1 are preprocessed and statistically screened, including: 1 Eliminate samples with incomplete data or data with significant errors; 2 Eliminate samples where the industry code does not exist; 3 Eliminate ST and *ST enterprise samples.
3. The statistical analysis method for digital transformation of automobile manufacturing according to claim 1, characterized in that: In the processing of the acquired data in step S1, the average value of the variable of the automobile enterprise in all years is selected to fill the missing values in the collected data, and the data is formatted.
4. The statistical analysis method for digital transformation of automobile manufacturing according to claim 1, characterized in that: In the S2 step, control variables are added based on the data, the frequency of occurrence of key words related to enterprise digital transformation is deeply analyzed, and these word frequency data are used to construct a comprehensive enterprise digital transformation index.
5. The statistical analysis method for digital transformation of automobile manufacturing according to claim 1, characterized in that: The comprehensive enterprise digital transformation index in step S2 can more comprehensively and accurately reflect the actual status and progress of enterprises in the digital transformation process. The specific steps are as follows: 1) Use Python programming technology to obtain financial data and analyze the company's growth trajectory and future planning clues; 2) Extract keywords for key areas and systematically measure and evaluate the digital transformation degree (DT) of enterprises through detailed indicator division; 3) Use Python technology to extract the annual report text data of listed companies, search for keywords and perform statistical processing of word frequency, and build an indicator system that comprehensively reflects the status of corporate digital transformation.
6. The statistical analysis method for digital transformation of automobile manufacturing according to claim 1, characterized in that: In order to improve the accuracy of the research in step S2, we added a series of control variables based on the statistical data of word frequency: enterprise resilience (EPR), goodwill (GA), cash flow level (CF), financial leverage (FZ), knowledge economy level (LOKE) and enterprise size (ES), and extracted and recorded the basic statistical characteristics of the introduced control variables, observed and summarized the changing characteristics of the statistical data.
7. The statistical analysis method for digital transformation of automobile manufacturing according to claim 1, characterized in that: Capture the changes in the company's sales (SV) relative to the entire industry, use the natural logarithm of the company's sales (SV) as a measure to reflect the increase or decrease in sales, eliminate the heteroscedasticity in the data, and more accurately reflect the company's true product market competitiveness (PMC) indicators in the product market.
8. The statistical analysis method for digital transformation of automobile manufacturing according to claim 1, characterized in that: In the S4 step, support vector regression (SVR) introduces a tolerance region and maximizes the interval of the tolerance region while minimizing the total loss to solve the regression model. It is divided into hard interval regression and soft interval regression. The interval of the tolerance region is maximized by artificially setting the ε deviation control. Sample points within the tolerance region are not included in the loss, and sample points outside the tolerance region need to be included in the loss.
9. The statistical analysis method for digital transformation of automobile manufacturing according to claim 8, characterized in that: For the sample anomalies and noise points in the partial tolerance region of step S4, it is difficult for the support vector hard margin regression model to determine the appropriate ε to ensure that most of the data are within the tolerance region, which seriously affects the prediction effect of the regression model. Therefore, soft margin regression is introduced.
