Large-screen data visualization method, system and equipment and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and particularly provides a data large-screen visualization method, system and device and a storage medium, and the method comprises the steps: capturing the field distribution characteristics of to-be-displayed data through a mixed density network; determining distribution type membership degrees corresponding to the field distribution characteristics through a fuzzy type membership degree calculation method; calculating the contribution degree of the distribution type membership degree to the display mode, and determining the type weight based on the contribution degree; calculating a decision value according to the distribution type membership degree, the contribution degree and the type weight, and determining an optimal display mode according to the decision value; the display mode comprises a statistical method and a chart type. According to the method, the internal mode of the data can be automatically identified, the optimal visualization scheme can be matched, and the problems that a traditional method is poor in adaptability and the display effect is one-sided are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence, and specifically relates to a method, system, device and storage medium for data large-screen visualization. Background Art

[0002] Existing large-screen visualization technologies mainly rely on manually preset rules or simple statistical analysis to select chart types. For example, based on field name matching or basic statistics (such as mean, variance), they are directly mapped to bar charts, line charts, etc. There are two major defects in such methods: one is that the complex characteristics of data distribution (such as multi-modal, skewed) are ignored, resulting in the chart being unable to accurately reflect the essence of the data; the other is that the rule configuration is rigid and it is difficult to adapt to mixed-type data (such as fields that simultaneously contain normal distribution and power-law distribution). Summary of the Invention

[0003] In view of the above deficiencies of the prior art, the present invention provides a method, system, device and storage medium for data large-screen visualization to solve the above technical problems.

[0004] In a first aspect, the present invention provides a method for data large-screen visualization, including: Using a mixture density network to capture the field distribution characteristics of the data to be displayed; Determining the distribution type membership corresponding to the field distribution characteristics through a fuzzy type membership calculation method; Calculating the contribution degree of the distribution type membership to the display mode, and determining the type weight based on the contribution degree; Calculating a decision value according to the distribution type membership, contribution degree and type weight, and determining the best display mode according to the decision value; The display mode includes statistical methods and chart types.

[0005] In an optional embodiment, using a mixture density network to capture the field distribution characteristics of the data to be displayed includes: Inputting the original data fields of the data to be displayed into the mixture density network to obtain the field distribution characteristics of the Gaussian mixture distribution, where the field distribution characteristics include mean, standard deviation and mixing coefficient; The mixture density network includes 5 layers of residual networks, and each layer of residual network contains 128 GRU units.

[0006] In an optional embodiment, determining the distribution type membership corresponding to the field distribution characteristics through a fuzzy type membership calculation method includes: The membership formula of the data to be displayed and the distribution type t is:

[0007] where K is the total number of Gaussian components into which the data to be displayed is decomposed; is the mean of the k-th Gaussian component; is the standard deviation of the k-th Gaussian component; is the mixing coefficient of the k-th Gaussian component; is the ideal mean of the t-th distribution type, which is a fixed value set according to expert experience; is the semantic vector of the field corresponding to the k-th Gaussian component and the standard semantic vector of the t-th distribution type similarity; is the dynamic adjustment factor.

[0008] In an alternative embodiment, calculating the contribution of the distribution type membership degree to the display mode and determining the type weight based on the contribution includes: Taking the membership degrees of the data to be displayed with each distribution type as features and the display mode as a label, and calculating the contribution of the membership degree to the display mode; Mapping the contribution to a type weight.

[0009] In an alternative embodiment, mapping the contribution to a type weight includes:

[0010] where r m ∈ {0, 1}, indicating whether the display mode m is selected; α represents the adjustment intensity factor; represents the new weight of the t-th distribution type, reflecting its importance; M represents the total number of display modes; ϕ t,m represents the original contribution of the t-th distribution type to the m-th display mode; ϕ t,m represents the new contribution of the t-th distribution type to the m-th display mode.

[0011] In an alternative embodiment, the method further includes: Updating the dynamic adjustment factor using the type weight:

[0012] where γ is the learning rate and ϵ is a small constant to prevent division by zero.

[0013] In an alternative embodiment, calculating a decision value based on the distribution type membership degree, contribution, and type weight, and determining the best display mode according to the decision value includes:

[0014] where DV m is the decision value for selecting the m-th display mode, is the membership degree of the data to be displayed with the t-th distribution type, is the weight of the t-th distribution type, is the contribution degree of the t-th distribution type to the m-th display mode; is the bias term; Filter out the maximum decision value, and output the display mode corresponding to the maximum decision value as the best display mode.

[0015] In a second aspect, the present invention provides a data dashboard visualization system, including: A feature extraction module, configured to capture the field distribution features of the data to be displayed by using a mixture density network; A type determination module, configured to determine the distribution type membership degree corresponding to the field distribution features by using a fuzzy type membership degree calculation method; A mode matching module, configured to calculate the contribution degree of the distribution type membership degree to the display mode, and determine the type weight based on the contribution degree; A mode determination module, configured to calculate a decision value according to the distribution type membership degree, the contribution degree, and the type weight, and determine the best display mode according to the decision value; The display mode includes a statistical method and a chart type.

[0016] In a third aspect, there is provided a device, including: A memory, configured to store a data dashboard visualization program; A processor, configured to implement the steps of the data dashboard visualization method provided in the first aspect when executing the data dashboard visualization program.

[0017] In a fourth aspect, there is provided a computer-readable storage medium, on which a data dashboard visualization program is stored. When the data dashboard visualization program is executed by a processor, the steps of the data dashboard visualization method provided in the first aspect are implemented.

[0018] The beneficial effects of the present invention are as follows. The data dashboard visualization method, system, device, and storage medium provided by the present invention can dynamically capture distribution features through a mixture density network, and combine fuzzy type membership degree calculation and contribution degree weighted decision-making to automatically identify the internal patterns of data and match the optimal visualization solution, solving the problems of poor adaptability and one-sided display effects of traditional methods.

[0019] In addition, the design principle of the present invention is reliable, the structure is simple, and it has a very wide application prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative efforts.

[0021] Figure 1 It is a schematic flowchart of the method according to an embodiment of the present invention.

[0022] Figure 2 It is a schematic block diagram of the system according to an embodiment of the present invention.

[0023] Figure 3 It is a schematic structural diagram of a device provided by an embodiment of the present invention. Detailed implementation manners

[0024] In order to enable those skilled in the art of the present technology to better understand the technical solutions in the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments, and are not intended to limit the present invention.

[0026] The data dashboard visualization method provided by the embodiments of the present invention is executed by a computer device. Correspondingly, the data dashboard visualization system runs in the computer device.

[0027] Figure 1 It is a schematic flowchart of the method according to an embodiment of the present invention. Among them, Figure 1 The execution subject can be a data dashboard visualization system. According to different requirements, the order of the steps in this flowchart can be changed, and some can be omitted.

[0028] Such as Figure 1 shown, the method includes: S1. Use a mixture density network to capture the field distribution characteristics of the data to be displayed; S2. Determine the distribution type membership degree corresponding to the field distribution characteristics through a fuzzy type membership degree calculation method; S3. Calculate the contribution degree of the distribution type membership degree to the display mode, and determine the type weight based on the contribution degree; S4. Calculate a decision value according to the distribution type membership degree, contribution degree and type weight, and determine the best display mode according to the decision value; the display mode includes statistical methods and chart types.

[0029] In an embodiment of the present invention, based on step S1, a possible embodiment will be given below to non - restrictively elaborate on its specific implementation.

[0030] Input the original data field of the data to be displayed into the mixture density network to obtain the field distribution characteristics of the Gaussian mixture distribution. The field distribution characteristics include the mean, standard deviation, and mixing coefficient. The mixture density network includes 5 - layer residual networks, and each layer of the residual network contains 128 GRU units.

[0031] Model the original data field through a Mixture Density Network (MDN) to obtain its field distribution characteristics of the Gaussian mixture distribution. Specifically, the input original data field of the data to be displayed first undergoes pre - processing, including operations such as standardization and missing value filling, to ensure the normality and usability of the data. The pre - processed data is used as the input of the mixture density network. Through the layer - by - layer calculation of the internal structure of the network, the parameters of the Gaussian mixture distribution, namely the mean μ, standard deviation σ, and mixing coefficient π, are finally output.

[0032] The mixture density network adopted in the present invention is based on a deep - learning architecture, and its core is a 5 - layer Residual Network. The introduction of the residual network effectively alleviates the problems of gradient disappearance and gradient explosion during the training process of the deep neural network, enabling the network to learn more complex feature representations. Each layer of the residual network contains 128 Gated Recurrent Units (GRUs). As a special type of Recurrent Neural Network (RNN) unit, the GRU selectively remembers and forgets information through a gating mechanism. Compared with traditional RNN units, it can more effectively handle long - term dependencies in sequence data. In this model, the GRU unit can capture potential time - series features or dependencies between data in the original data field.

[0033] During the network calculation process, the input data first enters the first - layer residual network. In each GRU unit, the data undergoes calculations of the reset gate r t and the update gate z t : r t =σ(W xr x t +W hr h t−1 +b r ) z t =σ(W xz x t +W hz h t−1 +b z ) Among them, x t is the input data at time t, h t−1 is the hidden state at the previous time, W xr , W hr , W xz , W hz are weight matrices, b r , b z are bias terms, and σ is the Sigmoid activation function. Through the reset gate and the update gate, the GRU cell can dynamically determine the information for retaining and updating the hidden state. The data processed by the first-layer residual network is passed to the next-layer residual network through a residual connection. The calculation formula of the residual connection is: h t = f(x t ) + x t Among them, f(x t ) is the calculation result of the remaining part in the residual network except the identity mapping. This structure enables the network to more easily optimize the objective function during the learning process and avoid the performance degradation problem caused by the network being too deep. After the layer-by-layer calculation of the 5-layer residual network, the feature vector output by the network is processed by a linear transformation and an activation function to obtain the parameters of the Gaussian mixture distribution. Assuming that the number of mixture components is K, for each data point, the probability density function of its Gaussian mixture distribution can be expressed as:

[0034] Among them, is the mixing coefficient of the k-th mixture component, is the probability density function of the Gaussian distribution with a mean of and a standard deviation of .

[0035] The training process of the network adopts the maximum likelihood estimation method, and the network parameters are optimized by minimizing the negative log-likelihood loss function.

[0036] In an embodiment of the present invention, based on step S2, a possible embodiment will be given below to non-restrictively elaborate on its specific implementation scheme.

[0037] The membership degree calculation formula of the data to be displayed and the distribution type t is:

[0038] Among them, K is the total number of Gaussian components into which the data to be displayed is decomposed; is the average value of the k-th Gaussian component; is the standard deviation of the k-th Gaussian component; is the mixing coefficient of the k-th Gaussian component; is the ideal mean of the t-th distribution type, which is a fixed value set based on expert experience; is the semantic vector of the field corresponding to the kth Gaussian component The standard semantic vector with the t-th distribution type similarity; is a dynamic adjustment factor.

[0039] The method for calculating the semantic similarity includes: encoding the data to be displayed by using an encoder to obtain a semantic vector, and then calculating the cosine similarity between the semantic vector of the data to be displayed and the standard semantic vector.

[0040] In an embodiment of the present invention, based on step S3, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner.

[0041] In the process of data processing and display mode optimization, in order to effectively mine the relationship between data features and display modes, this paper proposes an analysis method based on membership contribution calculation and type weight mapping. This method quantifies the influence of the membership of the data to be displayed and the distribution type on the display mode, providing a scientific basis for the selection of subsequent data visualization strategies.

[0042] S301. Take the membership of the data to be displayed and each distribution type as a feature, and the display mode as a label, and calculate the contribution of the membership to the display mode.

[0043]

[0044] (Feature SHAP value) is a scalar value used to quantify the contribution of a specific feature i to the model prediction results. This value is obtained through an improved causal intervention calculation method (different from the traditional conditional expectation method), which can more accurately explain the influence of features. Its technical core lies in the use of dynamic sampling strategies to generate feature subset combinations, and the Monte Carlo simulation is used to calculate the expected value of the prediction, thereby improving the credibility and stability of the explanation.

[0045] S (feature subset) represents the feature combination involved in the current analysis, and its generation process is based on the power set traversal of the full feature set F. The innovation of the patent is to optimize the traversal efficiency - through the dynamic sampling strategy, the feature combination with high probability association is preferentially selected, rather than exhaustively enumerating all possibilities. This strategy significantly reduces the computational complexity and is particularly suitable for scenarios where multimodal features (such as text, time series, and statistical features) are jointly analyzed.

[0046] F (Full set of features), representing all input feature sets defined in the system, whose sources include structured data field definitions, unstructured data parsing results, and multi-modal feature fusion outputs. In patent design, particular emphasis is placed on the ability to uniformly process heterogeneous features (such as numerical statistics, semantic embedding vectors, and time series pattern encodings), enabling different feature types to participate in decision-making collaboratively.

[0047] Factorial operation (S!), which is used here to calculate the combined weight coefficient of the feature subset S. Its theoretical basis stems from the Shapley value allocation principle in cooperative game theory. Through the factorial ratio of |S|! and (|F| - |S| - 1)!, the system can reasonably quantify the contribution weights of different feature combinations, ensuring the fairness of mathematical allocation. This mechanism is the mathematical guarantee for the traceability of explainable AI.

[0048] f x f(S) (model prediction expectation function), which outputs the model prediction expected value of the feature subset S. Its calculation process is realized through Monte Carlo simulation. The core innovation of the patent is to replace the traditional conditional expectation with causal intervention (do-calculus), avoiding the interference of spurious associations between features. For example, when analyzing the impact of "user age" on purchase behavior, the system will force the setting of age values (do operation), rather than simply counting the conditional probability of age distribution, so as to more accurately capture causal relationships.

[0049] Feature subset sampling strategy:

[0050] λ (temperature parameter), which is a positive scalar parameter used to control the concentration degree of the sampling distribution. The default empirical value is 0.7, which is optimized and determined on the validation set through grid search. Its core role is to balance the relationship between exploration (extensive sampling) and exploitation (focusing on high-probability regions). When λ increases, the sampling distribution tends to be uniform, enhancing the exploration of rare feature combinations; when λ decreases, the sampling is concentrated in the high-probability region, improving the calculation efficiency.

[0051] q(S) (current model prediction distribution), representing the prediction probability distribution of the current model for the feature subset S, which is directly obtained through the Softmax value output by the model. For example, in a classification task, if the prediction probability of the model for a subset S is 0.8, then q(S) = 0.8. This parameter dynamically reflects the real-time confidence of the model in feature combinations and is the basis for calculating sampling weights.

[0052] p base(Base Distribution), which is a predefined base distribution, usually based on the occurrence frequency of the statistical features combination of the training set, and Laplace smoothing is used to avoid the zero-probability problem. For example, if a certain feature combination appears 100 times in the training set and the total sample size is 100,000, then its base probability is (100 + 1) / (100,000 + total number of feature combinations). The base distribution provides stability guarantee to prevent sampling bias caused by short-term fluctuations of the model.

[0053] KL divergence (D_KL(q||p_base)) is used to measure the difference between the current model's predicted distribution q(S) and the base distribution p_base. The calculation formula is:

[0054] A larger KL value indicates that the current model behavior significantly deviates from the historical pattern, which may imply data distribution drift or model overfitting.

[0055] Dynamic adjustment mechanism: When data distribution drift is detected (JS divergence > 0.3):

[0056] User feedback-driven adjustment:

[0057] The sampling probability of the feature subset S is calculated by the following formula: The feature subset sampling strategy realizes the following functions: KL penalty: The greater the difference between the model's predicted distribution q(S) and the base distribution, the lower the sampling probability, which suppresses abnormal fluctuations.

[0058] Temperature control: λ adjusts the sensitivity to the difference. The higher λ is, the greater the allowed distribution deviation.

[0059] Model prediction difference calculation:

[0060] f x (S) represents the function output, and its technical meaning is: it represents the predicted expected value of the model for the current data sample when the given feature subset is S. The data source is: calculated by the Monte Carlo simulation method, that is, by randomly sampling the feature values outside the feature subset S multiple times, simulating the model prediction results in different scenarios, and taking the average value.

[0061] Causal intervention (do-calculus) is adopted instead of the traditional conditional expectation calculation. The traditional method only calculates the conditional probability based on the existing data distribution, while causal intervention eliminates the spurious association between features by actively intervening in the feature values (such as forcing a certain feature to be a specific value), so as to more truly reflect the causal impact of features on the prediction results.

[0062] f x (S) Calculation method: Monte Carlo simulation is a numerical calculation method based on random sampling. When calculating f x (S), the specific steps are as follows: Fix the feature subset S: Keep the feature values in S unchanged.

[0063] Randomly perturb the non-S features: Randomly sample the features that do not belong to S (i.e., F∖S) and replace them with other sample values in the training set or random values that conform to the distribution.

[0064] Predict multiple times and take the average: Repeat the above perturbation process N times (e.g., N = 1000), and get the prediction result y of the model each time i , and finally calculate the average value.

[0065] The traditional conditional expectation E[y∣X S =x S may be biased due to the correlation between features (such as confounding factors). For example, if features A and B are correlated, the traditional method cannot distinguish the direct causal effect of A on the result and the indirect effect through B. And causal intervention measures the independent influence directly by forcing X S =x S (such as do(X S =x S )) to cut off the dependence relationship between S and other features.

[0066] S302. Map the contribution degree to a type weight.

[0067]

[0068] where r m ∈{0,1}, indicating whether the display mode m is selected; α represents the adjustment intensity factor; represents the new weight of the t-th distribution type, reflecting its importance; M represents the total number of display modes; ϕ t,m represents the original contribution degree of the t-th distribution type to the m-th display mode; ϕ t,m represents the new contribution degree of the t-th distribution type to the m-th display mode.

[0069] S303. Adjust the membership degree calculation formula using the type weight.

[0070] Update the dynamic adjustment factor using the type weight:

[0071] where γ is the learning rate and ϵ is the anti-zero constant.

[0072] In an embodiment of the present invention, based on step S4, a possible embodiment will be given below to non - restrictively elaborate on its specific implementation scheme.

[0073]

[0074] Among them, DV m is the decision value for selecting the m - th display mode, is the membership degree of the data to be displayed with respect to the t - th distribution type, is the weight of the t - th distribution type, is the contribution degree of the t - th distribution type to the m - th display mode; is the bias term.

[0075] Among them, the type weight w t As a global decision regulator, the global dimension w t represents the overall importance weight of type t (such as continuous type, categorical type, time - series type) in the current system environment. As a global regulation parameter, it determines the priority ranking of different types in the decision - making process. For example, when the system detects that the data has strong time - series characteristics, the time - series weight w_time - series will increase significantly, so that time - related statistical methods (such as moving average) obtain higher selection priority.

[0076] In the calculation of the decision value, w t acts as a multiplicative factor to scale the type influence, and its value range is [0,1]. The higher the weight of the type, the stronger the guiding role of the type in the final decision.

[0077] The SHAP contribution value ϕ t,m is a local causal interpreter, and the local dimension ϕ t,m quantifies the causal contribution degree of type t to the specific display mode m (such as the combination of summation, mean, quantile and chart type). Its core role is to explain why a certain type of feature tends to select a specific method under the current data distribution. For example, the high contribution value of the categorical weight to "frequency statistics" (ϕ_categorical,frequency = 0.9) indicates that the discrete characteristics of the categorical field are naturally suitable for count analysis.

[0078] ϕt,m ∈ [−1,1]. A positive value indicates that type t has a positive causal impact on display mode m, and a negative value indicates inhibition.

[0079] The type weight w t and the SHAP contribution value ϕ t,m drive the decision from two dimensions of system - level environment perception and data - level causal reasoning respectively. The product fusion design of the two solves the contradiction between global rigidity and local over - fitting in traditional methods.

[0080] Filter out the maximum decision value and output the display mode corresponding to the maximum decision value as the optimal display mode.

[0081] In some embodiments, the data dashboard visualization system may include multiple functional modules composed of computer program segments. The computer programs of each program segment in the data dashboard visualization system may be stored in the memory of the computer device and executed by at least one processor to perform the functions of data dashboard visualization (see Figure 1 description).

[0082] In this embodiment, the data dashboard visualization system can be divided into multiple functional modules according to the functions it performs, as Figure 2 shown. The module referred to in the present invention means a series of computer program segments that can be executed by at least one processor and can complete fixed functions, and are stored in the memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0083] Feature extraction module, configured to capture the field distribution features of the data to be displayed by using a mixture density network; Type determination module, configured to determine the distribution type membership corresponding to the field distribution features by using a fuzzy type membership calculation method; Mode matching module, configured to calculate the contribution degree of the distribution type membership to the display mode and determine the type weight based on the contribution degree; Mode determination module, configured to calculate a decision value according to the distribution type membership, contribution degree, and type weight, and determine the optimal display mode according to the decision value; The display mode includes statistical methods and chart types.

[0084] Figure 3 The data dashboard visualization method provided in the embodiments of the present application can be applied to devices. Those skilled in the art can understand that the device structure involved in the embodiments of the present invention does not constitute a limitation on the device. The device may include more or fewer components than shown, or combine certain components, or have different component arrangements. In the embodiments of the present invention, the device includes but is not limited to laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of the present application described herein and / or claimed.

[0085] Among them, the device 300 may include: a processor 310, a memory 320, and a communication unit 330. These components communicate through one or more buses. Those skilled in the art can understand that the structure of the server shown in the figure does not constitute a limitation on the present invention. It can be a bus structure, a star structure, and may also include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0086] Among them, the memory 320 can be used to store the execution instructions of the processor 310. The memory 320 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk. When the execution instructions in the memory 320 are executed by the processor 310, the device 300 can execute some or all of the steps in the above method embodiments.

[0087] The processor 310 is the control center of the storage device, connecting various parts of the entire electronic device through various interfaces and lines. By running or executing the software programs and / or modules stored in the memory 320, and calling the data stored in the memory, it executes various functions of the electronic device and / or processes data. The processor may be composed of an integrated circuit (IC). For example, it may be composed of a single packaged IC, or may be composed of multiple packaged ICs with the same or different functions connected together. For example, the processor 310 may only include a central processing unit (CPU). In the embodiment of the present invention, the CPU may be a single arithmetic core or may include multiple arithmetic cores.

[0088] The communication unit 330 is used to establish a communication channel, so that the storage device can communicate with other devices. It receives user data sent by other devices or sends user data to other devices.

[0089] The present invention also provides a computer storage medium. Among them, the computer storage medium can store a program, and when the program is executed, it can include some or all of the steps in the embodiments provided by the present invention. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0090] Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solutions in the embodiments of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc., which are various media that can store program codes, and includes several instructions for causing a computer device (which may be a personal computer, a server, or a second device, a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0091] For the same or similar parts among the various embodiments in this specification, reference can be made to each other. In particular, for the device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the descriptions in the method embodiments.

[0092] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of systems or modules can be in electrical, mechanical, or other forms.

[0093] The modules described as separate components may or may not be physically separated. The components shown as modules may or may not be physical modules, that is, they can be located in one place, or they can be distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0094] In addition, in each embodiment of the present invention, the various functional modules can be integrated in a processing module, or each module can exist physically alone, or two or more modules can be integrated in one module.

[0095] Although the present invention has been described in detail by reference to the accompanying drawings and in conjunction with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, those of ordinary skill in the art can make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions should all be within the scope of the present invention / Any person skilled in the art within the technical scope disclosed by the present invention can easily conceive of changes or substitutions, which should all be covered by the protection scope of the present invention.

Claims

1. A method for visualizing a large data screen, characterized in that, including: capturing the field distribution characteristics of the data to be displayed by using a mixture density network; determining the distribution type membership corresponding to the field distribution characteristics through a fuzzy type membership calculation method; calculating the contribution degree of the distribution type membership to the display mode, and determining the type weight based on the contribution degree; calculating a decision value according to the distribution type membership, the contribution degree and the type weight, and determining the best display mode according to the decision value; the display mode includes statistical methods and chart types.

2. The method according to claim 1, wherein Capturing the field distribution characteristics of the data to be displayed by using a mixture density network includes: inputting the original data fields of the data to be displayed into the mixture density network to obtain the field distribution characteristics of the Gaussian mixture distribution, where the field distribution characteristics include the mean, the standard deviation and the mixing coefficient; the mixture density network includes 5 layers of residual networks, and each layer of residual network contains 128 GRU units.

3. The method according to claim 1, wherein Determining the distribution type membership corresponding to the field distribution characteristics through a fuzzy type membership calculation method includes: the membership formula of the data to be displayed and the distribution type t is: Where K is the total number of Gaussian components into which the data to be displayed is decomposed; is the mean of the k-th Gaussian component; is the standard deviation of the k-th Gaussian component; is the mixing coefficient of the k-th Gaussian component; is the ideal mean of the t-th distribution type, which is a fixed value set according to expert experience; is the semantic vector of the field corresponding to the k-th Gaussian component and the standard semantic vector of the t-th distribution type similarity; is the dynamic adjustment factor.

4. The method according to claim 3, characterized in that, Calculating the contribution degree of the distribution type membership to the display mode, and determining the type weight based on the contribution degree includes: taking the membership of the data to be displayed and each distribution type as features, taking the display mode as a label, and calculating the contribution degree of the membership to the display mode; mapping the contribution degree to the type weight.

5. The method according to claim 4, characterized in that, Mapping the contribution degree to the type weight includes: where r m ∈ {0, 1}, indicating whether the display mode m is selected; α represents the adjustment intensity factor; represents the new weight of the t-th distribution type, reflecting its importance; M represents the total number of display modes; ϕ t,m represents the original contribution degree of the t-th distribution type to the m-th display mode; ϕ t,m represents the new contribution degree of the t-th distribution type to the m-th display mode.

6. The method according to claim 5, characterized in that, the method further includes: updating the dynamic adjustment factor by using the type weight: where γ is the learning rate and ϵ is the anti-zero constant.

7. The method according to claim 1, wherein Calculating a decision value according to the distribution type membership, the contribution degree and the type weight, and determining the best display mode according to the decision value includes: Among them, DV m is the decision value for selecting the m-th display mode, is the membership degree of the data to be displayed and the t-th distribution type, is the weight of the t-th distribution type, is the contribution degree of the t-th distribution type to the m-th display mode; is the bias term; screening out the maximum decision value, and outputting the display mode corresponding to the maximum decision value as the best display mode.

8. A data dashboard visualization system, characterized in that, including: a feature extraction module for capturing the field distribution characteristics of the data to be displayed by using a mixture density network; a type determination module for determining the distribution type membership corresponding to the field distribution characteristics through a fuzzy type membership calculation method; a mode matching module for calculating the contribution degree of the distribution type membership to the display mode, and determining the type weight based on the contribution degree; a mode determination module for calculating a decision value according to the distribution type membership, the contribution degree and the type weight, and determining the best display mode according to the decision value; the display mode includes statistical methods and chart types.

9. A data large-screen visualization device, characterized in that, including: a memory for storing a data large screen visualization program; a processor for implementing the steps of the data large screen visualization method as described in any one of claims 1-7 when executing the data large screen visualization program.

10. A computer-readable storage medium storing a computer program, characterized in that, The data large screen visualization program is stored on the readable storage medium, and when the data large screen visualization program is executed by the processor, the steps of the data large screen visualization method as described in any one of claims 1-7 are implemented.

Images