Big data analysis technology in regional economic market planning and enterprise market value prediction

2.1 Development status abroad

Along with the growth of regional economic market planning, the regional economic market planning approach also evolves continuously. Planning for foreign regional economic markets has a lengthy history [5]. The lengthy study procedure has consistently improved and enhanced the planning method, which is currently quite sophisticated.

The design of the Custer coal mine in England from 1922 to 1923 and the urban regional planning of New York in the United States in 1929 are two examples of how foreign regional economic market planning is based on urban planning and industrial and mining location planning [6]. The planning scope, however, is modest and straightforward for this time frame. Most regional economic market planning techniques until the 1940s were relatively straightforward, primarily utilizing industrial location and geography methods [7]. The importance of cities’ function and position in the local economy grew by the 1940s due to the growth of capitalism, increased industrialization, and urbanization. The planning process meets numerous challenges [8]. Greater London Regional Plan was first created by the United Kingdom in 1944, after which industrialized nations, the former Soviet Union, and Eastern European nations generally implemented a large-scale regional economic market plan [9]. The input-output method and linear programming have become quite popular, and the regional economic market planning method has advanced rapidly. The input-output method application offers a potent tool for interregional trade and research of regional industrial structure; linear programming offers convenience for resource allocation and sensible use.

Regional economic market planning has advanced to a new level since the 1960s, with the study of people, society, environment, and resources caused by capitalist industrialization and urbanization [10]. A simple economy gives way to a complete planning of five factors in the regional economy: the economy, people, society, environment, and resources. During this time, the regional economic market planning approach has advanced to a new stage of development to respond to the complexity of regional economic comprehensive planning. Regional economic market planning techniques have gradually evolved in the West away from conventional techniques like qualitative description, zoning, and analysis and toward techniques like systems engineering, grey control system, AHP method, system dynamics model, and multi-objective decision planning [11]. Another significant turning point in the innovation of planning methods and tools occurred around 1990 when several regions began using decision support system technology in their regional economic market planning efforts [12]. With these new technologies, planners and decision-makers can more effectively address various planning-related issues, increase job productivity, lessen the subjective influence of human variables, and focus on planning’s quantitative issues [13]. Regional economic market planning is now a multi-disciplinary, all-encompassing process rather than a straightforward approach, including a few disciplines. For instance, countries use computers as the primary tool for regional planning and fully utilize remote sensing, geographic information systems (GIS), global positioning system, and other high-tech and mathematical modeling technologies.

On the one hand, this makes it convenient for the public to quickly and easily participate in planning and understand the planning results, but on the other hand, it alters the planning process. In the past, the planning process was primarily based on qualitative research, and the area was the subject of quantitative study. Fuzzy science techniques for uncertain factor analysis and mathematical models and model systems are frequently employed in mathematical applications. Western regional economic market planning techniques are progressing towards an all-encompassing model of quantification and modeling.

2.2 Domestic development status

The application research for the regional economy planning approach is still in its infancy in my country. My nation once used regional planning based on joint processing plants from 1953 to 1957 to assist with developing new industrial zones and cities. Only straightforward geographic, mathematical, and locational techniques were applied [14]. The central government decided to develop land planning as we entered the 1980s. Most provinces, cities, and districts have planned land in succession and obtained specific results [15]. Input-output tables were first collected by Shanxi Province in 1982. Subsequently, many other provinces, municipalities, and districts followed Shanxi’s lead. Since then, Xinjiang has created strategies for its social and economic development [16]. In order to create development market plans in the Simo area of Yunnan and Jining County of Kunming City, Sun et al. [17] established a complete set of basic index model systems and a dynamic rolling model system of induced indexes in 1990. As a result, the designated strategic goals and plans have a comparatively high level of performance, high adaptability, and precision. In 1997, Wang et al. [18] proposed a Pareto optimum algorithm for identifying answers by fusing the positive eigenvector approach with network technology, providing a regional economic network model. The theory, guiding principles, scope, and regional economic market planning techniques were studied and elaborated upon by Lin et al. [19] in 2020. They talked about the concepts, modeling procedures, and principles of applying multi-objective optimization techniques to challenges in regional economic market planning and incorporated some of their ideas into examples of specific applications that were examined. In 2020, Hung et al. [20] reviewed the use of GIS in county urban system planning and discussed how it might be used in regional planning research to manage data, analyze data, support decision-making, and articulate planning conclusions. In the study by You and Wu [21], it is suggested that the use of remote sensing and GIS technology be made more common in regional planning in the new era, resulting in the digitization of the entire regional planning process, from field research, data collection, processing, and computational simulation to planning and mapping, implementation, and supervision. In 2007, Radwan et al. [22] developed a comprehensive index evaluation method appropriate for the western region using fuzzy comprehensive evaluation for the region’s planning. The same year, regional economic market planning used a hybrid analytical hierarchy process and grey relational decision-making.

3.1 Analysis of regional economic market planning

The recursive convolutional neural network (RCNN) model is chosen in this study to analyze the direction of market planning in regional economic time series (RETS-RCNN). It is introduced as economic news articles and technical indicators.

As shown in Figure 3,

where b → ∈ R R is the bias term. Equation (1) is applied to the sentence window to generate the feature map Q,

Figure 3

TS-RNN.

To introduce nonlinearity into the model, the rectifier linear unit is used as the activation function,

Since this layer involves many parameters, the model is prone to overfitting, so a Dropout regularization technique with probability p = 0.5 is used.

3.2 Setting and estimation of a regional econometric model

The spatial dependency of economic growth between regions must serve as the foundation for measuring diffusion backflow and market area effect between regions. Spatial economics studies commonly utilize binary matrices consisting of zeros and ones to examine spatial effects. A weight matrix is employed to gauge the spatial interdependence or connectivity among countries, indicating proximity and influencing the extent of spatial spillovers between them. To evaluate whether there is spatial correlation across regions and to establish the properties of spatial correlation, we first build a spatial econometric model. At the same time, we thoroughly chose the model using the fixed-effect LR test. The spatial lag model is used in this study for the test results, to examine the geographic relationship between the economics of a province and a county. The comparison model used in this study is

where α is the constant term, ρ and β are the coefficients to be estimated, W is the spatial weight matrix, δ is the individual fixed, u is the time fixed, and ε is the random error term, which obeys a normal distribution.

Through the spatial weight matrix, the spatial correlation between areas primarily affects each region. Unique patterns, clusters, and the influence of adjacent regions can be identified by breaking down spatial relationships into discrete elements. This method improves understanding of how spatial factors lead to differences between various places. As a result, the spatial weight matrix is primarily decomposed for the spatial decomposition. Because a given province’s regional economic development has created a center-secondary-periphery, the interaction of the same level area, the diffusion and return of the central area to the secondary area, the diffusion and return of the central area to the peripheral area, the diffusion and return of the secondary area to the peripheral area, and the secondary area are the primary divisions of the spatial effect between the center and the periphery. Encouraging regional development can foster economic growth in surrounding areas. However, it may also result in a return of economic activities to the central area, impacting the overall regional economic landscape. The dynamic interplay between diffusion and return shapes regional economic development. Primary and peripheral regions on central regions and peripheral regions on secondary regions are the effects of market area. Therefore, this study divides three different levels of regions into L, M, and S, and the economic output of the corresponding region is Y L , Y M , Y S on average. Therefore, the decomposition of the spatial weight matrix can be in the form of a block matrix,

Then, the total effect of spatial spillover between regions of different levels can be expressed as

At the same time, the dummy variables representing the three types of areas affected by the spillover effect are multiplied by the above formula to determine the measure of the source and receiver of the spillover effect. If the L-type region is the receptor, it should be W ML Y L I , and so on, when estimated in the model; if the i set observation of the sample is the L-type region, then W LL Y L L ≠ 0 , W ML Y L L ≠ 0 W SL Y L L ≠ 0 other spatial autocorrelations. The variables are all 0, and so on. Therefore, the Spatial Autoregressive (SAR) model based on spatial decomposition to measure the directional spatial correlation between regions of different levels can be determined as

Including dummy variables in the model makes it easier to identify and quantify spillover effects and distinguish various groups or areas. This method improves the assessment of the interactions between adjacent units and considers the data’s heterogeneity, which makes the SAR framework better.

Therefore, ρ LL , ρ MM , ρ SS represent the spatial spillover effect between cities at the same level, ρ LM , ρ LS , ρ MS represent the diffusion and reflux effect of the central area on the secondary area, the central area on the peripheral area, and the secondary area on the peripheral area, and ρ ML , ρ SL , ρ SM represent the secondary area, respectively. The impacts of market areas on central regions, central regions on peripheral regions, and peripheral regions on secondary regions.

3.3 Enterprise market value forecast

According to the above study, an enterprise’s market value comprises its current and potential value. Enterprise value represents a company’s overall value, similar to a balance sheet, while market value reflects its stock value. Both are used in business valuation, with enterprise value indicating current worth and market value showing both current and potential future value. As a result, this essay will also assess the two components of value independently. The potential value V2 is evaluated using actual options, whereas the current value V1 is evaluated using the conventional income technique. The assessment approach is displayed in Figure 4.

1. Building a discounted cash flow model investment in project R&D, acquisition, reorganization, and scale-up activities, as well as a fast increase in operating income. The company’s growth potential will stabilize at a mostly fixed level, and at this point, the company will enter a stable period due to the internet industry’s “first dominance” and the Matthew effect.

Figure 4

Composition of the basic model for enterprise market value assessment.

The binary tree model is based on a discrete process, and it uses dynamic programming to handle the interaction between options, underlying assets, and decision-making. The binary tree model, a discrete option valuation method, breaks time into intervals to simulate asset price changes. Unlike continuous models such as Black–Scholes, it is adept at handling complexities like American-style options and variable volatility. Its flexibility suits various option types and market conditions. This method more accurately simulates the calculation of option value. Although the binary tree model simplifies the B-S pricing model, it is more appropriate for evaluating and pricing particular investment projects since its calculation involves a substantial number of information sets, such as the projected value of options at each time node. It supports discrete time intervals and adapts well to diverse asset price changes, making it ideal for intricate investment projects like American-style options or those with irregular cash flows. This flexibility enables accurate valuation by considering various scenarios at each step. Although the B-S model’s derivation procedure is rather complex, acquiring its fundamental parameters does not necessitate a significant amount of data collecting and processing,

where C 0 is the current option value; S 0 is the existing value of the underlying asset; X is the strike price; e is the natural logarithmic base; r c is the continuously compounded risk-free rate of return (years); t is the maturity time (years); and σ is the underlying asset return volatility.

In calculating an enterprise’s potential value, the most crucial step is to determine the value of each parameter, such as S 0 , X t S 0 σ , etc. Currently, the underlying asset of the real option is an enterprise, and the specific meanings of the above parameters are as follows:

1. C 0 is the potential option value of the company. The overall value of the enterprise includes the existing asset value and the potential value, and the potential value is evaluated through the option pricing formula.

2. X is the execution price, which can usually be regarded as the investment cost in the evaluation of enterprise market value.

3. t is the exercise period of the natural option. The evaluation of the market value of the enterprise refers to the period from the evaluation base date to the expiration date of the natural option.

4. r c is the risk-free interest rate under continuous compounding, which can be replaced by the treasury bond maturity yield that is the same or similar to the option expiration date.

5. σ is the volatility of the underlying asset return. For companies, it usually refers to the volatility of the stock price.

Since it has been established that the big data analysis portfolio evaluation model is used to assess the market value of businesses, the fundamental model needs to be upgraded. As a result, the improved grey prediction approach forms the foundation for further model optimization. This method is appropriate for predicting “poor information, few samples” systems because it can distinguish differences in the development trends of different system components through correlation analysis, system law mining, and the establishment of differential equations. On the one hand, the macro environment that businesses operate in is unstable, and the future revenue of businesses has a high level of uncertainty. On the other hand, the grey forecasting approach may use known data to anticipate the future revenue of businesses with a high level of uncertainty. We enhance prediction accuracy by integrating sophisticated algorithms, utilizing historical data with a focus on recent trends, and adjusting dynamically to evolving conditions.

On the other hand, most businesses are still in the startup or development phase, and there is a need for more publicly available data, making it challenging to use techniques like time series forecasting. However, the grey forecasting method has fewer requirements for historical data, which can help businesses overcome their lack of historical data; prepare for the future by analyzing market trends, customer preferences, and competitor actions; seek advice from industry experts, stay updated on emerging trends, and gather customer feedback; and monitor competitors for strategic insights, address risks, and use internal data for performance evaluation and predictive modeling.

In terms of volatility forecasting, this study uses forecasted volatility to replace actual volatility, and the grey forecasting method also applies to forecasting corporate volatility.

1. The following takes the GM(1,1) model to predict the operating income as an example to illustrate the steps of using the model.

3.4 Model accuracy evaluation

1. The residual error test calculates the relative and absolute errors and observes whether the error changes are stable by calculating the residual error.

2. The correlation degree test is calculated, and the test is passed when the value is more significant than 0.5. The metric assesses the model’s ability to capture data patterns, with a strong correlation indicating accurate predictions. This efficient test reliably evaluates the model’s performance by quantifying the strength and direction of the relationship.

3. Then, calculate the small error probability as

According to the above research, the conventional GM(1,1) model can forecast businesses’ cash flow and volatility, but it also has several drawbacks. The model, for instance, works well with data series that exhibits traits of mild volatility and exponential change. However, as indicated in Table 2, most businesses are in the startup or growth stages, with multiple development and investment stages.

This study will correct the nonlinear residual generated by the original model. Polynomial regression analysis is a critical regression analysis method. Since polynomials can approximate any function, it can solve related problems in nonlinear regression. When revising the GM(1,1) model, polynomial regression can be introduced to perform polynomial regression on the residual between the predicted sequence X ⌢ ( 0 ) ( K ) and the initial sequence X ( 0 ) ( K ) (i.e., e ( 0 ) ( k ) ), and the original residual sequence e ( 0 ) ( k ) performs fitting prediction, and then uses the obtained prediction residual value e ˆ ( 0 ) ( k ) to correct the prediction sequence X ˆ ( 0 ) ( k ) . The polynomial model enhances prediction accuracy by capturing curvature in the data, unlike the linear GM(1,1) model. It provides a flexible representation of underlying patterns, improving the model’s ability to approximate nonlinear trends and make accurate predictions. At this time, the prediction error due to the characteristics of the original sequence can be corrected to improve the prediction accuracy further.

The revised GM(1,1) model can increase forecasting accuracy by rewriting the nonlinear residual and is more suited for enterprise data forecasting. The grey forecasting theory can handle the forecasting problem of the system with uncertainty. Its ability to thrive with limited data, adeptness in handling uncertainty, and incorporation of historical and expert insights make it well-suited for unpredictable economic environments. Its flexibility and efficacy in addressing uncertain aspects make it a valuable tool for predicting revenue in turbulent business conditions. The enterprise fluctuation is comparatively severe, and the new model can partially correct the prediction inaccuracy brought on by this nonlinear fluctuation through adaptive simulation. The enhanced grey forecasting model has significant implications for business valuation when taken as a whole. Figure 5 below illustrates how the enterprise market value evaluation model was improved. The market value of an enterprise is determined using methods like discounted cash flow analysis, comparable company analysis, and precedent transactions. Factors influencing market value include revenue and profit growth, market share, competitive position, industry trends, management quality, and economic conditions.

Figure 5

Improvement roadmap for the basic model of enterprise market value evaluation.

4.1 Data sources

Logarithmic standardization is applied to the data in this research, taken from the statistical yearbooks of prefecture-level cities in X Province from 2001 to 2019. This section analyses data from 2001 to 2019 as a whole.

4.2 Measurement of regional development difference indicators

Theil entropy, which measures the degree of regional development imbalance and distributes the share of regional imbalance across and within groups, can also be used to scale the Theil index. Based on the assumption that the entire economy, represented by individuals, will be divided into groups, the loss index used to measure regional imbalances is calculated.

By decomposing the Theil index, it can be divided into inter-group gap and intra-group gap.

The Gini coefficient (G) is an indicator used to measure and judge regional income or wealth distribution inequality. It is constructed according to the Lorenz curve, and the international warning value of the Gini coefficient is 0.4.

By studying the Theil index decomposition and the calculation of the Gini coefficient, this study finds that the current regional economic pattern in X Province presents some new phenomena as follows.

In X Province, the ongoing increase of the regional development disparity has been somewhat curbed. Since 2008, the X region has lagged behind the X region regarding GDP growth rate (Figure 6). We measure the imbalanced trend of economic development in X Province using the widely used Theil index, Gini coefficient, and Atkinson index. The change trend is essentially the same for all three, despite some variances in numerical measurement. Before 2006, the regional economic difference in X Province showed a rising tendency. After 2006, the regional disparity steadily declined and became more petite yearly (Figure 7). However, it cannot be denied that the enormous absolute level disparity between the X Province’s east, west, and north regions and the A region continues to be a significant issue for the province’s economic development. The spatial lag model studies spatial patterns in economic factors, incorporating the impact of neighboring areas through weighted relationships. It assesses how economic conditions in one location affect nearby areas, revealing and quantifying spatial trends and interactions in economic data across provinces and counties.

Figure 6

Comparison of GDP growth rates in the A and X’s east, west, and north.

Figure 7

The overall difference measure index of X Province.

There is a clear gap between the central cities of Guangzhou and Shenzhen and other regions, and the internal development of the X region is differentiated.

The gap between the core area of the A and the non-A area is the primary manifestation of the regional gap in X Province. The decrease in disparities across groups primarily reflected the decline in regional differences. While intra-group differences were broadly steady, changes in group differences and the overall index followed the same pattern. Further, we looked at the gap within the group and discovered that the faster the region developed, the bigger the distance within the group was, as shown in Figures 8 and 9. The rate of economic development is relatively rapid. However, because of the increased regional competition, it has also led to differences in economic development rates within the group. The distinctions within groups grew as a result of this. The non-A region has always had a shaky base for economic growth, and there has never been much industrial advancement [19,23]. The per capita GDP distribution is relatively uniform, and the development disparity between X’s east, west, and north sections is insignificant. As X’s internal divisions between its east, west, and north sections strengthen, so do they.

Figure 8

Theil index decomposition of regional differences in X Province.

Figure 9

Intra-group difference index by region.

Using internet businesses as an example, those defined narrowly are more indicative of the sector. The essential details regarding specifically categorized internet companies listed in my country. A total of 187 unicorn companies, with a combined market value of 11.2 trillion yuan, were among the 296 domestically listed companies with a market size of more than 3.39 trillion yuan as of the end of December 2019 (excluding the first category of companies). Internet-listed businesses increased from 2008 to 2019, and their market shares also increased.