Regional development quality evaluation method and system fusing multi-dimensional data

Abstract

The invention relates to the technical field of data processing, in particular to a regional development quality evaluation method and system fusing multi-dimensional data, and the method comprises the steps: obtaining and preprocessing multi-source data related to regional economic coordinated development, so as to extract different data features and form a standardized feature data set; fusing the plurality of feature data sets by using a VGG convolutional neural network to obtain fused index features; clustering the fusion index features through a fuzzy C-means clustering algorithm, extracting a clustering center and calculating a membership matrix to construct an evaluation model; and judging and evaluating the newly acquired multi-source data by using the evaluation model so as to obtain an evaluation result of regional economic coordinated development. According to the scheme, full-process automation from data acquisition, index construction to intelligent evaluation is realized, and scientificity, objectivity and efficiency of an evaluation system are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology. More specifically, this invention relates to a method and system for evaluating regional development quality by integrating multi-dimensional data. Background Technology

[0002] In recent years, rapid economic development has brought about various problems, prompting us to question traditional theories of economic growth. Therefore, while reflecting on the complex relationships beyond the "quantity" of economic development, we need to pay special attention to harmony between humanity and nature, harmony among people, and emphasize coordinated development between regions.

[0003] Coordinated regional economic development is an important indicator of the economic health and sustainability of a country or region. Its assessment involves multiple dimensions, including economic, social, resource, and environmental aspects, and is a complex systems engineering project. To scientifically and objectively evaluate the level of coordinated regional economic development, a comprehensive evaluation indicator system is typically required. The construction of this system often relies on the comprehensive analysis and processing of massive amounts of diverse and heterogeneous data from various sources (such as statistical yearbooks, policy texts, news images, and satellite remote sensing data).

[0004] For a long time, due to the many factors affecting the coordinated development of regional economies, the selection of indicator systems for measuring the coordinated development of regional economies may face the problem of being unable to select the best and resulting in an overemphasis on indicators. This has brought certain difficulties to the subsequent use of indicator systems to conduct assessments of the degree of coordination of regional economies.

[0005] However, a prominent technical challenge exists in constructing an evaluation indicator system using multi-source data: how to automatically and efficiently screen and integrate core indicators that accurately reflect the state of coordinated development from a vast and high-dimensional set of potential indicators. Traditional methods often rely on expert experience for manual screening or simple linear weighting, which is not only highly subjective and inefficient, but also prone to leading to an bloated indicator system and information redundancy when dealing with massive amounts of data. This makes it difficult to reveal the deep nonlinear relationships between indicators, thus affecting the accuracy and practicality of the evaluation model.

[0006] Therefore, how to solve the problems of poor accuracy and low practicality in the current assessment of regional economic development quality is one of the key research directions. Summary of the Invention

[0007] To address the aforementioned technical problems of poor accuracy and low practicality in current assessments of regional economic development quality, this invention provides solutions in the following aspects.

[0008] In a first aspect, the present invention provides a method for evaluating the quality of regional development by integrating multi-dimensional data, comprising: acquiring and preprocessing multi-source data related to coordinated regional economic development to extract different data features and form a standardized feature dataset; using a VGG convolutional neural network to fuse multiple feature datasets to obtain fused index features; clustering the fused index features using a fuzzy C-means clustering algorithm, extracting cluster centers and calculating a membership matrix to construct an evaluation model; and using the evaluation model to discriminate and evaluate the newly acquired multi-source data to obtain the evaluation results of coordinated regional economic development.

[0009] In one embodiment, the multi-source data includes: macroeconomic policy texts, statistical yearbooks, human resources data, fixed asset investment data, fiscal revenue and expenditure data, environmental assessment reports, natural resource inventories, development data of various production sectors, and infrastructure construction and operation data.

[0010] In one embodiment, preprocessing multi-source data related to coordinated regional economic development includes: processing text data using the TF-IDF algorithm to extract text features.

[0011] In one embodiment, preprocessing multi-source data related to coordinated regional economic development includes: framing and Fourier transforming speech data to generate spectrogram data.

[0012] In one embodiment, fusing multiple feature datasets using a VGG convolutional neural network includes: concatenating the multiple feature datasets into a high-dimensional feature matrix, and inputting the high-dimensional feature matrix into a pre-trained VGG convolutional neural network for convolution processing.

[0013] In one embodiment, the calculation formula for convolution is:

[0014]

[0015] in, Let M represent the i-th point in the t-th feature layer, f() represent an activation function, and M... t Indicates the number of sliding windows. This represents the j-th unit in the (t-1)-th input layer. This represents the j-th unit in the i-th convolutional layer of the t-th feature layer. This represents the i-th offset in the t-th convolutional layer.

[0016] In one embodiment, the fuzziness parameter m of the fuzzy C-means clustering algorithm takes values ​​in the range of [1.5, 2.5].

[0017] In one embodiment, using an evaluation model to discriminate and evaluate newly acquired multi-source data includes: preprocessing and fusing the newly acquired multi-source data to obtain an index vector to be evaluated; calculating the similarity or membership degree between the index vector and each cluster center; and taking the category corresponding to the cluster center with the highest similarity or membership degree as the evaluation result.

[0018] In one embodiment, the operation process of the fuzzy C-means clustering algorithm includes: initializing the membership matrix; calculating the cluster centers based on the current membership matrix; calculating the Euclidean distance between all index features and each cluster center; updating the membership matrix according to the Euclidean distance until the algorithm converges.

[0019] In a second aspect, the present invention also provides a regional development quality evaluation system that integrates multi-dimensional data. The system includes: a processor; and a memory storing computer program instructions that, when executed by the processor, implement a regional development quality evaluation method that integrates multi-dimensional data as described in one or more embodiments of the first aspect.

[0020] The beneficial effects of this invention are as follows: By automatically collecting multi-source information and utilizing VGG neural network technology to automatically extract and fuse feature indicators from different modalities, and then combining this with fuzzy C-means clustering algorithm to establish an evaluation model, this invention achieves a fully automated and intelligent process from data acquisition and indicator extraction to model building and evaluation. This method not only overcomes the limitations of manual indicator selection and can handle complex multi-dimensional data, but also objectively mines the deep correlations between data through machine learning algorithms, thereby establishing a more scientific, objective, and comprehensive evaluation indicator system for coordinated regional economic development, significantly improving the accuracy and efficiency of the evaluation.

[0021] Furthermore, this invention clarifies the specific dimensions of information collection, covering multiple levels including macro, meso, and micro. This multi-dimensional and targeted data collection ensures the comprehensiveness and representativeness of the input data sources, enabling the subsequently established evaluation model to more accurately reflect the true state of regional economic development. It avoids evaluation biases caused by single or missing data dimensions, enhancing the practical guiding significance and reliability of the evaluation results.

[0022] Furthermore, this invention uses a deep network structure to better capture complex, high-order intrinsic patterns and correlations in economic data, thereby improving the effectiveness of fusion indicator extraction and the performance of the final evaluation model, enabling it to make more accurate judgments than shallow models. Attached Figure Description

[0023] Figure 1This is a flowchart illustrating a method for evaluating regional development quality by integrating multi-dimensional data according to an embodiment of the present invention;

[0024] Figure 2 This is a structural diagram illustrating a regional development quality evaluation system that integrates multi-dimensional data according to an embodiment of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0027] This invention aims to address the technical problem in existing technologies where the numerous factors influencing coordinated regional economic development make it difficult to select optimal evaluation indicators and consequently, to conduct effective assessments. The core innovation of this invention lies in constructing a regional economic coordinated development evaluation system that integrates artificial intelligence technology. It abandons the traditional method of relying on human experience to select indicators and innovatively employs VGG convolutional neural networks to process and fuse heterogeneous data from multiple sources, including text, voice, and images, to automatically extract deep feature indicators. Combined with fuzzy C-means clustering, a multi-dimensional evaluation model is objectively established, achieving full automation from data collection and indicator construction to intelligent evaluation. This significantly improves the scientific rigor, objectivity, and efficiency of the evaluation system, providing precise decision support for regional economic development.

[0028] The technical solutions in the embodiments of the present invention will be clearly and completely described below.

[0029] The present invention provides a method for evaluating regional development quality by integrating multi-dimensional data. The overall process can be summarized as follows:

[0030] In step S101: acquire and preprocess multi-source data related to coordinated regional economic development to extract different data features and form a standardized feature dataset.

[0031] In one embodiment, the first step is to comprehensively collect multi-source, heterogeneous data related to the coordinated development of the regional economy. This data forms the foundation for subsequent indicator extraction and model building. As a preferred approach, data collection can be carried out around the following dimensions:

[0032] (1) Considering the overall layout of national economic development, collect and analyze data such as macroeconomic policy documents and statistical yearbooks related to the status and role of the target region's economy in the national economic system.

[0033] (2) Collect data that reflects the speed and scale of regional economic development and its compatibility with local carrying capacity, such as human resources data, fixed asset investment data, and fiscal revenue and expenditure data.

[0034] (3) Collect data such as environmental assessment reports and natural resource lists related to regional economic development plans and planning and design, in order to assess resource utilization efficiency and environmental protection level.

[0035] (4) Collect development data of various production sectors (such as agriculture, industry and service industry) in the region, as well as construction and operation data of infrastructure such as energy, transportation, communication, medical and health care, culture and education, in order to analyze the degree of coordinated development between industrial structure and infrastructure.

[0036] After acquiring the aforementioned multi-source data, preprocessing is required to extract implicit indicator information and form a standardized dataset. Specifically:

[0037] For text data, such as policy documents and research reports, the TF-IDF (Term Frequency-Inverse Document Frequency) algorithm can be used for preprocessing. This algorithm evaluates the importance of a word to a document in a document set or corpus by calculating the product of term frequency and inverse document frequency, thereby transforming unstructured text into structured text feature vectors.

[0038] For speech data, such as recordings of expert interviews, the audio signal is first segmented into frames to extract each audio frame. Then, a Fourier transform is performed on each audio frame to obtain its spectral value. Finally, spectrogram data is generated based on the spectral values. A spectrogram is a two-dimensional image where the horizontal axis represents time, the vertical axis represents frequency, and the intensity of the color indicates the energy level at that frequency, providing a visual representation of the frequency domain characteristics of the speech signal.

[0039] For image data, such as remote sensing images and infrastructure photographs, a uniform scaling and grayscale conversion are first performed to reduce computational complexity and eliminate interference from color information. For example, all images can be uniformly scaled to 224×224 pixels.

[0040] Through the above preprocessing, text data features, speech data features (spectral graph training set), and image data features (image training set) can be obtained respectively, laying the foundation for subsequent fusion feature extraction.

[0041] In step S102: VGG convolutional neural network is used to fuse multiple feature datasets to obtain fused index features.

[0042] After obtaining preliminary indicator information and datasets, this embodiment utilizes VGG convolutional neural network technology to perform deep fusion and abstract feature extraction of these multi-source features.

[0043] As a preferred embodiment, the VGG convolutional neural network includes several convolutional layers, several pooling layers, and a fully connected layer. Each convolutional layer is followed by a corresponding pooling layer. Through multiple layers of convolution and pooling operations, the network can extract abstract features from the input data layer by layer, from low to high level.

[0044] In this embodiment, the text data features, speech data features, and image data features obtained in step S101 are first fused, for example, by concatenating them into a high-dimensional feature matrix. Then, the fused feature matrix is ​​input into a pre-trained VGG convolutional neural network to extract the fused features.

[0045] In the VGG convolutional neural network, the image convolution calculation process can be represented by the following formula. This formula is based on simulating the response mechanism of visual cortical cells to local receptive fields. By sliding the convolution kernel on the input feature map, feature detection and extraction are achieved.

[0046]

[0047] in, Let M represent the i-th point in the t-th feature layer, f() represent an activation function, and M... t Indicates the number of sliding windows. This represents the j-th unit in the (t-1)-th input layer. This represents the j-th unit in the i-th convolutional layer of the t-th feature layer. This represents the i-th offset in the t-th convolutional layer.

[0048] According to the above formula, the output features... It consists of all input features from the previous layer After convolution kernel Weighted summation, with bias added. Then, it is obtained through the activation function f(). When the input features... When a change occurs, if its corresponding weight A positive and large value will significantly affect the output. The value reflects how the neural network learns to adjust weights to capture key patterns in the input data. In this embodiment, through the multi-layer nonlinear transformation of the VGG network, deeper and more discriminative fusion index features can be extracted from the fused data.

[0049] In step S103: The fusion index features are clustered using the fuzzy C-means clustering algorithm, the cluster centers are extracted and the membership matrix is ​​calculated to construct the evaluation model.

[0050] After extracting the features of the fusion indicators, this embodiment uses the Fuzzy C-Means (FCM) clustering algorithm to perform cluster analysis on these feature data, aiming to explore the intrinsic relationship between the indicator features and the state of coordinated regional economic development, and finally construct an evaluation model.

[0051] The process of constructing the evaluation model specifically includes:

[0052] Initialization Unit: Let the feature set of integrated indicators for coordinated regional economic development be X = {x1, x2, ..., x...} n}, which contains n index vectors. First, a membership matrix U needs to be randomly initialized, where the elements u ij Represents the j-th index vector x j The membership degree of the i-th cluster center. This matrix needs to satisfy the following constraints:

[0053]

[0054] Where c is the preset number of clusters. This formula indicates that the sum of the membership degrees of any index vector to all clusters must be 1.

[0055] Cluster center calculation unit: The purpose of this step is to calculate c cluster centers C = {c1, c2, ..., c3} based on the current membership matrix U. c Its construction is based on finding a center point that can represent each category. This center point should be the weighted average position of all sample points within that category. The calculation formula is as follows:

[0056]

[0057] Among them, c i x represents the i-th cluster center; j It is the j-th index vector; u ij It is x j For c i The membership degree; m is a hyperparameter representing the degree of ambiguity. From this formula, we can see that the cluster center c... i It is the vector of all indicators x jThe weighted average is calculated, with each weight being the m-th power of its membership degree. When an index vector x... j Membership degree y of category i ij When it increases, it affects the center c of that category. i The "pulling" effect will be enhanced, making c i To x j The direction of movement. As a preferred option, the ambiguity parameter m ranges from [1.5, 2.5]. When the value of m is too low, the clustering result tends to be a hard partition; when the value of m is too high, the ambiguity of the clustering result is too large, and the category boundaries are unclear. Therefore, setting m to 2 can usually obtain stable and effective clustering results in most application scenarios.

[0058] Value Function Calculation Unit: To evaluate the quality of the current clustering, the value function J needs to be calculated. It is constructed based on the sum of weighted distances from all sample points to their respective cluster centers. A good clustering result should minimize intra-cluster distances. The calculation formula is as follows:

[0059]

[0060] Where, d ij =||c i -x j || represents the i-th cluster center c i With the j-th index vector x j The Euclidean distance between them. The smaller the value function J, the closer the sample point is to the cluster center with high membership, and the better the clustering effect. The iterative process of the algorithm is the process of minimizing J.

[0061] Membership Matrix Update Unit: Based on the newly calculated cluster centers, the membership degree of each index vector to each category needs to be updated. This is based on the principle that the closer a sample point is to a cluster center, the greater the probability that it belongs to that class. The update formula is as follows:

[0062]

[0063] From this formula, we can see that the index vector x j For cluster center c i membership degree u ij , and its distance d from the center ij Inversely proportional. When the distance d ij When decreasing, the term in the denominator (d) ij d kj ) decreases, thus making u ij The value of increases, which aligns with the intuitive logic that the closer the distance, the higher the membership degree.

[0064] The algorithm iterates through steps 102 to 104 until the value function J converges to a value less than a preset threshold or the maximum number of iterations is reached, at which point the algorithm stops. The resulting cluster centers C and the final membership matrix U together constitute the evaluation model of this invention. For example, the c cluster centers can be categorized according to their characteristic meanings as "highly coordinated development," "moderately coordinated development," "uncoordinated development," etc.

[0065] In step S104: The newly acquired multi-source data is evaluated and assessed using an evaluation model to obtain the evaluation results of coordinated regional economic development.

[0066] Once the assessment model is built, it can be used to identify and evaluate newly acquired multi-source information and data, and obtain the assessment results of regional economic coordinated development.

[0067] Specifically, the process involves: acquiring multi-source economic information and data for the region to be processed, and performing preprocessing and feature extraction according to steps S101 and S102 to obtain the indicator vector to be evaluated. Subsequently, this indicator vector is input into the constructed evaluation model to calculate its similarity or membership degree with each cluster center (e.g., "coordinated development" and "uncoordinated development" clusters). The cluster category corresponding to the indicator vector with the highest similarity or membership degree is taken as the final evaluation result for the region to be evaluated.

[0068] For example, stratification can be achieved based on the probability corresponding to each score. For instance, regions with a membership degree greater than 60% could be classified as "coordinated regional economic development," while those with a membership degree less than 60% could be classified as "uncoordinated regional economic development." Personalized and professional solutions and development suggestions can then be provided for different stratification results.

[0069] In summary, this invention, through the collection of multi-source data and the use of VGG networks and fuzzy C-means clustering algorithms, scientifically constructs an evaluation index system and model for coordinated regional economic development. This approach not only simplifies the indicator selection process but also achieves accurate assessment and dynamic tracking of the quality of regional economic development through a data-driven approach, effectively avoiding the interference of subjective factors in traditional methods and making the evaluation results more objective and accurate.

[0070] Figure 2 This is a schematic diagram illustrating the structural block of a regional development quality evaluation system that integrates multi-dimensional data according to this embodiment.

[0071] This invention also provides a regional development quality evaluation system that integrates multi-dimensional data. As shown in the figure, the system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the regional development quality evaluation method that integrates multi-dimensional data according to the first aspect of this invention.

[0072] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and therefore will not be described in detail here.

[0073] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented using computer-readable / executable instructions that can be stored or otherwise maintained by such a computer-readable medium.

[0074] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise explicitly specified.

[0075] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A method for evaluating regional development quality by fusing multi-dimensional data, characterized in that, The method comprises the following steps: acquiring and preprocessing multi-source data related to regional economic coordination development to extract different data features and form a standardized feature data set; fusing multiple feature data sets by using a VGG convolutional neural network to obtain fused index features; clustering the fused index features by using a fuzzy C-means clustering algorithm, extracting cluster centers and calculating a membership matrix to construct an evaluation model; using the evaluation model to discriminate and evaluate newly acquired multi-source data to obtain an evaluation result of regional economic coordination development. 2.The regional development quality evaluation method of fusing multi-dimensional data according to claim 1, characterized in that, The multi-source data includes macro policy texts, statistical yearbooks, human resource data, fixed asset investment data, financial revenue and expenditure data, environmental assessment reports, natural resource lists, development data of various production sectors, and construction and operation data of infrastructure. 3.The regional development quality evaluation method of fusing multi-dimensional data according to claim 1, characterized in that, The preprocessing of the multi-source data related to regional economic coordination development includes processing text data by using a TF-IDF algorithm to extract text features.

4. The method according to claim 1, wherein, The preprocessing of the multi-source data related to regional economic coordination development includes framing and Fourier transforming voice data to generate spectrogram data.

5. The method according to claim 1, wherein, The fusion of multiple feature data sets by using a VGG convolutional neural network includes: splicing the multiple feature data sets into a high-dimensional feature matrix, and inputting the high-dimensional feature matrix into a pre-trained VGG convolutional neural network for convolution processing.

6. The method according to claim 5, wherein, The calculation formula of the convolution processing is: wherein, represents the i-th point in the t-th feature map layer, f() represents an activation function, M t represents the number of sliding windows, represents the j-th unit in the t-1 input layer, represents the j-th unit in the i-th convolutional layer in the t-th feature map layer, represents the i-th offset in the t-th convolutional layer.

7. The method according to claim 1, wherein, The fuzzy degree parameter m of the fuzzy C-means clustering algorithm ranges from 1.5 to 2.

5. 8.The method of claim 1, wherein, The discrimination and evaluation of the newly acquired multi-source data by using the evaluation model includes: preprocessing and fusing the newly acquired multi-source data to obtain an index vector to be evaluated; calculating the similarity or membership of the index vector to each cluster center; the class corresponding to the cluster center with the highest similarity or membership is taken as the evaluation result.

9. The method of claim 1, wherein the method is characterized by, The operation process of the fuzzy C-means clustering algorithm includes: initializing the membership matrix; calculating the cluster centers based on the current membership matrix; calculating the Euclidean distances between all index features and the cluster centers; updating the membership matrix according to the Euclidean distances until the algorithm converges.

10. A system for evaluating regional development quality by fusing multi-dimensional data, characterized by, The system comprises: a processor; a memory storing computer program instructions that, when executed by the processor, implement the regional development quality evaluation method of fusing multi-dimensional data according to any one of claims 1-9.

Images