Signal intersection public opinion prediction method and system based on integrated algorithm prediction model

Abstract

The invention provides a signalized intersection public opinion prediction method and system based on an integrated algorithm prediction model, and relates to the technical field of public opinion prediction. The method mainly adopts an integrated algorithm prediction model with a semi-naive Bayes algorithm, a logistic regression algorithm and a classification decision tree as a base learner; according to the method, the public opinions of the urban signalized intersection are analyzed and predicted through the strong classifier, and the public opinion prediction result can be accurately improved by improving the generalization ability and performance of the model.

Description

Technical Field

[0001] The present invention relates to the technical field of public opinion prediction, and in particular to a signal intersection public opinion prediction and system based on an integrated algorithm prediction model. Background Art

[0002] With the continuous advancement of urbanization and economic development, the national transportation demand continues to increase, and the demand for residents' travel quality is also constantly improving. In recent years, with the further integration of transportation, the Internet, and big data, major traffic emergencies are increasingly exposed and spread through the Internet, and new media platforms such as social media have become the primary platform for the rapid spread of traffic public opinion. Traffic public opinion analysis and monitoring have become an important part of solving traffic problems, and are also an important development direction for the close integration of transportation, social public opinion, and urban governance in the future.

[0003] Traffic public opinion is an exposure of current traffic problems and a feedback on the performance of the current traffic system. It is rich in important traffic-related information. At present, the analysis of urban traffic public opinion is still in its infancy. There is a lack of in-depth analysis of urban traffic public opinion data, and traffic public opinion information has not been effectively used to assist in urban traffic governance. On the other hand, the rapid development of artificial intelligence provides new ideas for the analysis and resolution of traffic problems. It is crucial to rationally utilize traffic big data and combine it with intelligent methods to discover hidden dangers in urban traffic, conduct proactive intervention, and improve the quality of urban traffic services. At this stage, relevant scholars mostly use the naive Bayes method to carry out public opinion prediction, but the model is not accurate enough and the influencing factors are not considered enough. Summary of the invention

[0004] The purpose of the present invention is to provide a signal intersection public opinion prediction and system based on an integrated algorithm prediction model. The method mainly adopts a semi-naive Bayes algorithm, a logistic regression algorithm and a classification decision tree as an integrated algorithm prediction model of base learners, and combines them into a strong classifier through weighted combination. The strong classifier is used to analyze and predict the public opinion of urban signal intersections, and the generalization ability and performance of the model can be improved to accurately improve the public opinion prediction results.

[0005] In order to solve the above technical problems, the present invention adopts the following solutions:

[0006] A signal intersection public opinion prediction method based on an integrated algorithm prediction model, the method specifically comprises the following steps:

[0007] S1: Obtain the characteristic data of all signalized intersections and public opinion information related to signalized intersections;

[0008] S2: preprocessing the characteristic data and public opinion information of the signalized intersection, and establishing a signalized intersection public opinion data set based on the preprocessed characteristic data and public opinion information of the signalized intersection;

[0009] The signalized intersection public opinion data set includes several sample subsets, each sample subset takes a public opinion information as a sample, and collects characteristic data of relevant signalized intersections as sample features of the sample;

[0010] S3: Substitute the sample features in the sample subset into the integrated algorithm prediction model to perform sample weighted prediction in an iterative manner, and wait for the best prediction sample to be obtained after the iteration is completed.

[0011] Further, in S3, the integrated algorithm prediction model includes a first base learner, a second base learner, and a third base learner; the first base learner, the second base learner, and the third base learner are all used to perform sample prediction on the current sample features according to the weights of the current sample features.

[0012] Furthermore, in S3, in each iteration process, the sample features in the sample subset are sequentially predicted by the first base learner, the second base learner, and the third base learner to obtain sample prediction results, and the predicted samples in the sample prediction results obtained by the third base learner are used as the best prediction samples.

[0013] Furthermore, in S3, the sample weighted prediction process is specifically as follows:

[0014] In the process of iterating the first base learner, the second base learner, and the third base learner, the weight of the current sample feature in the current base learner is obtained, so that the current base learner performs sample prediction on the sample feature with the weight of the current sample feature, thereby obtaining the sample prediction result of the current base learner, and then the weight of the sample feature in the next base learner is weighted according to the sample prediction result of the current base learner, so that the next base learner performs sample prediction on the sample feature with the weight of the weighted sample feature, and so on.

[0015] Furthermore, the sample prediction result includes a prediction sample, and the process of weighting the weight of the sample feature in the next base learner according to the sample prediction result of the current base learner is specifically as follows:

[0016] The predicted sample obtained by the current base learner is compared with the corresponding sample, and whether the predicted sample is an incorrectly predicted sample is determined based on the comparison result. If so, the weight of the sample feature in the current base learner is increased to obtain the increased weight, and the weight of the sample feature in the next base learner is updated based on the increased weight.

[0017] Furthermore, the first base learner, the second base learner, and the third base learner all have base learner weights, and the process of obtaining the weight of the current sample feature in the current base learner is specifically as follows:

[0018] The weights of the sample features in the current base learner are updated according to the base learner weights of the current base learner itself and the weights increased according to the weights of the sample features in the previous base learner, so as to obtain the weights of the sample features in the current base learner.

[0019] Furthermore, in S2, the process of preprocessing the characteristic data and public opinion information of the signalized intersection includes the following steps:

[0020] SA1: Perform data cleaning and data desensitization on the characteristic data and public opinion information of signalized intersections, and remove false public opinion information based on personal information and intersection information in the public opinion information;

[0021] SA2: Determine the public opinion category of each public opinion information and mark it according to the public opinion category;

[0022] SA3: Determine the intersection feature data contained in the intersection information in each public opinion information based on the intersection features.

[0023] Furthermore, in S2, each sample subset takes a piece of public opinion information as a sample, marks the public opinion category of the sample through the public opinion information, and uses the intersection feature data contained in the intersection information in the public opinion information as the sample feature of the sample.

[0024] Furthermore, the first base learner uses a super-parent dependency estimation model as a base learner, the second base learner uses a logistic regression model as a base learner, and the third base learner uses a decision tree model as a base learner.

[0025] A signal intersection public opinion prediction system based on an integrated algorithm prediction model, comprising:

[0026] Data collection module: obtains characteristic data of all signalized intersections and public opinion information related to signalized intersections;

[0027] Data processing module: pre-process the characteristic data and public opinion information of the signalized intersection, and establish a signalized intersection public opinion data set based on the pre-processed characteristic data and public opinion information of the signalized intersection;

[0028] The signalized intersection public opinion data set includes several sample subsets, each sample subset takes a public opinion information as a sample, and collects characteristic data of relevant signalized intersections as sample features of the sample;

[0029] Sample weighted prediction module: Substitute the sample features in the sample subset into the integrated algorithm prediction model to perform sample weighted prediction in an iterative manner, and wait for the best prediction sample to be obtained after the iteration is completed.

[0030] Beneficial effects of the present invention:

[0031] The present invention provides a signal intersection public opinion prediction and system based on an integrated algorithm prediction model. The method mainly adopts a semi-naive Bayes algorithm, a logistic regression algorithm and a classification decision tree as an integrated algorithm prediction model of base learners, and combines them into a strong classifier through weighting. The strong classifier is used to analyze and predict the public opinion of urban signal intersections, and the generalization ability and performance of the model can be improved to accurately improve the public opinion prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flow chart of public opinion prediction for a signal intersection based on an integrated algorithm prediction model in Example 1 of the present invention.

[0033] Figure 2 Schematic diagram of the composition of the integrated algorithm prediction model in Example 1 of the present invention.

[0034] Figure 3 This is a schematic diagram of the process of performing sample weighted prediction by the base learner in Example 1 of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. The following description of at least one exemplary embodiment is actually only illustrative and is by no means intended to limit the present invention and its application or use. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] The relative arrangement of components and steps, the numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.

[0037] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0038] Additionally, descriptions of well-known structures, functions, and configurations may be omitted for clarity and conciseness.One of ordinary skill in the art will recognize that various changes and modifications may be made to the examples described herein without departing from the spirit and scope of the present disclosure.

[0039] Technologies, methods, and apparatus known to ordinary technicians in the relevant field may not be discussed in detail, but where appropriate, such technologies, methods, and apparatus should be considered part of the authorization specification.

[0040] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0041] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments:

[0042] Example 1

[0043] like Figure 1 As shown, a signal intersection public opinion prediction method based on an integrated algorithm prediction model comprises the following steps:

[0044] S1: Obtain the characteristic data of all signalized intersections and public opinion information related to signalized intersections;

[0045] S2: preprocessing the characteristic data and public opinion information of the signalized intersection, and establishing a signalized intersection public opinion data set based on the preprocessed characteristic data and public opinion information of the signalized intersection;

[0046] The signalized intersection public opinion data set includes several sample subsets, each sample subset takes a public opinion information as a sample, and collects characteristic data of relevant signalized intersections as sample features of the sample;

[0047] S3: Substitute the sample features in the sample subset into the integrated algorithm prediction model to perform sample weighted prediction in an iterative manner, and wait for the best prediction sample to be obtained after the iteration is completed.

[0048] Preferably, in S3, the integrated algorithm prediction model includes a first base learner, a second base learner, and a third base learner; the first base learner, the second base learner, and the third base learner are all used to perform sample prediction on the current sample features according to the weights of the current sample features.

[0049] Preferably, in S3, in each iteration process, the sample features in the sample subset are sequentially predicted by the first base learner, the second base learner, and the third base learner to obtain sample prediction results, and the predicted samples in the sample prediction results obtained by the third base learner are used as the best prediction samples.

[0050] Preferably, in S3, the process of sample weighted prediction is specifically as follows:

[0051] In the process of iterating the first base learner, the second base learner, and the third base learner, the weight of the current sample feature in the current base learner is obtained, so that the current base learner performs sample prediction on the sample feature with the weight of the current sample feature, thereby obtaining the sample prediction result of the current base learner, and then the weight of the sample feature in the next base learner is weighted according to the sample prediction result of the current base learner, so that the next base learner performs sample prediction on the sample feature with the weight of the weighted sample feature, and so on.

[0052] Preferably, the sample prediction result includes a prediction sample, and the process of weighting the weight of the sample feature in the next base learner according to the sample prediction result of the current base learner is specifically as follows:

[0053] The predicted sample obtained by the current base learner is compared with the corresponding sample, and whether the predicted sample is an incorrectly predicted sample is determined based on the comparison result. If so, the weight of the sample feature in the current base learner is increased to obtain the increased weight, and the weight of the sample feature in the next base learner is updated based on the increased weight.

[0054] Preferably, the first base learner, the second base learner, and the third base learner all have base learner weights, and the process of obtaining the weight of the current sample feature in the current base learner is specifically as follows:

[0055] The weights of the sample features in the current base learner are updated according to the base learner weights of the current base learner itself and the weights increased according to the weights of the sample features in the previous base learner, so as to obtain the weights of the sample features in the current base learner.

[0056] Preferably, in S2, the process of preprocessing the characteristic data and public opinion information of the signalized intersection includes the following steps:

[0057] SA1: Perform data cleaning and data desensitization on the characteristic data and public opinion information of signalized intersections, and remove false public opinion information based on personal information and intersection information in the public opinion information;

[0058] SA2: Determine the public opinion category of each public opinion information and mark it according to the public opinion category;

[0059] SA3: Determine the intersection feature data contained in the intersection information in each public opinion information based on the intersection features.

[0060] Preferably, in S2, each sample subset takes one piece of public opinion information as a sample, marks the public opinion category of the sample through the public opinion information, and uses the intersection feature data contained in the intersection information in the public opinion information as the sample feature of the sample.

[0061] Preferably, the first base learner uses a super-parent dependency estimation model as a base learner, the second base learner uses a logistic regression model as a base learner, and the third base learner uses a decision tree model as a base learner.

[0062] Based on the above principles, the present invention is further described:

[0063] In order to accurately improve the results of public opinion prediction, the present invention provides a public opinion prediction for signal intersections based on an integrated algorithm prediction model. The method mainly adopts a semi-naive Bayes algorithm, a logistic regression algorithm and a classification decision tree as an integrated algorithm prediction model of base learners, and combines them into a strong classifier through weighted combination, and uses the strong classifier to analyze and predict the public opinion at urban signal intersections.

[0064] Specifically, in this embodiment, the following steps are used to predict the key public opinion types at different signal-controlled intersections, thereby serving as a guide for the intersection timing plan and operation organization plan, and realizing active prevention of public opinion and improvement of service quality:

[0065] Step 1: Collect public opinion information on issues related to signal intersection timing in the survey area from various platforms such as the Internet and hotlines, and identify intersection feature data contained in the intersection information in the public opinion information, that is, the feature labels of the intersection.

[0066] Step 2: Based on the collected public opinion data, data preprocessing is performed, including data cleaning and data desensitization. Untrue public opinion records need to be eliminated, and personal information (name, mobile phone number, user name, etc.) and specific intersection names need to be desensitized.

[0067] Step 3: Determine and mark the public opinion category of each public opinion information based on the public opinion category. Specifically, the public opinion can be divided into the following six categories based on the length of time for crossing the intersection, the release method, traffic safety, facilities and equipment, etc.:

[0068] c1: Complaints about pedestrian crossing time (insufficient / too long);

[0069] c2: Complaints related to motor vehicle time (insufficient / excessive / unreasonable reallocation of time);

[0070] c3: Complaint about unreasonable plan (unreasonable release method);

[0071] c4: Coordinate and link related complaints;

[0072] c5: Complaints related to safety issues (conflicts between people and vehicles / conflicts between traffic);

[0073] c6: Complaints about facilities and equipment related issues (whether supporting facilities, faults, and manual controls have been optimized).

[0074] Step 4: Determine the intersection feature data contained in each intersection information in public opinion information according to the intersection features, that is, select the feature labels of the intersections in the public opinion information, which should include the intersection's own features, environmental features, etc. Specifically, the feature labels of the intersections can be divided into the following six types:

[0075] x1: intersection level (level 1, 2, or 3);

[0076] x2: intersection type (cross, T, special shape);

[0077] x3: intersection congestion index;

[0078] x4: vehicle speed at the intersection;

[0079] x5: Construction intersection / section (yes, no);

[0080] x6: Comprehensive scenarios (protecting intersections, focusing on intersections, business districts, schools, scenic spots, and hospitals).

[0081] Step 5: Perform data standardization processing, and use the public opinion information as a subset of the normalized data to form a signalized intersection public opinion data set. The signalized intersection public opinion data set is expressed as:

[0082] T=(x1,y1),(x2,y2),…,(x N ,y N ), where x i =(x 1 ,x 2 ,…,x n ), is the sample feature, which is composed of a series of independent feature dimensions, namely, intersection feature data; y i =c k , where k = 1, 2, ... K, y is the sample category, that is, the public opinion category, belonging to c1 to c k Any value in .

[0083] Step 6: Substitute the sample features in the sample subset into the integrated algorithm prediction model to perform sample weighted prediction in an iterative manner, and wait for the iteration to end to obtain the best prediction sample. In addition, the integrated algorithm prediction model can explore the relationship between intersection characteristics and key public opinions to identify key public opinions at specific intersections. It can be trained by randomly selecting base learners and their corresponding parameters in iterations to enhance the accuracy and robustness of public opinion prediction. The final strong classifier integrates the prediction results of each weak classifier through a weighted voting mechanism to improve the overall classification performance of the integrated model, thereby realizing the prediction of key public opinions based on intersection characteristics.

[0084] Specifically, Figure 2As shown, the integrated algorithm prediction model includes the following three base learners:

[0085] 1. Super Parent Dependency Estimation (SPODE)

[0086] Super Parent Dependency Estimator (SPODE) is an improved Bayesian classifier that simplifies the calculation of conditional probability by introducing a super parent node. Its core formula is as follows:

[0087] Class prior probability:

[0088]

[0089] Where N is the number of samples, I(·) is the indicator function, which takes the value of 1 when the condition is true and 0 otherwise. Y represents the sample category.

[0090] Feature conditional probability:

[0091]

[0092] Among them, x i represents the feature vector of the i-th sample, x sp is the super parent node feature vector, v and u are feature x i and x sp The value of α is the smoothing parameter, V is the feature x i The value of .

[0093] Classification probability:

[0094]

[0095] Among them, i represents the sample number, x is the feature vector, y represents the sample category, and x sp is the super parent node feature vector, P(x i |y,x sp ) is given a class label y and a super parent node feature x sp Under the condition that i The conditional probability of .

[0096] 2. Logistic Regression

[0097] Logistic regression is a linear classification model that predicts category probabilities by learning the parameters of a linear function. Its core formula is as follows:

[0098] Class Probability:

[0099]

[0100] Among them, w T are model parameters and X is the feature vector.

[0101] Loss function (logarithmic loss):

[0102]

[0103] 3. Decision Tree

[0104] The decision tree recursively splits the data set and builds a tree structure for classification. Its core formula is as follows:

[0105] Information Gain:

[0106]

[0107] Among them, IG(D,A) is the information gain of feature A on data set D, H(D) is the entropy of data set D, and D v is a subset of feature A whose value is v.

[0108] entropy:

[0109]

[0110] Among them, j represents the category number, k is the number of categories, and p j is the probability of class j.

[0111] Based on the above three base learners, weighted processing is performed, and the predicted samples obtained by the current base learner are compared with the corresponding samples. According to the comparison results, it is determined whether the predicted samples are wrongly predicted samples. If so, the weights of the sample features in the current base learner are increased to obtain the increased weights, and the weights of the sample features in the next base learner are updated according to the increased weights.

[0112] Specifically, the sample weight is updated as follows:

[0113] In each round of iteration, the sample weight D is updated according to the classification results of the base learner:

[0114] D t+1 (i) = D t (t)exp(-α t y i h t (x i ))

[0115] Among them, i represents the sample number, t represents the base learner number, and x i represents the feature vector of the i-th sample, α t is the base learner h t The weight of y i is the true label of the i-th sample, h t (xi ) is the base learner h t The prediction for the i-th sample. t Represents the input sample weight when the t-th learner predicts.

[0116] Among them, the base learner weight is:

[0117]

[0118] Among them, e t is the error rate of the t-th base learner,

[0119] In summary, if Figure 3 As shown, each time a base learner is used for sample prediction, the weights of the sample features in the current base learner need to be updated based on the base learner weights of the current base learner itself and the weights of the sample features in the previous base learner that are increased.

[0120] Based on the above iterative process, the collected data can also be pre-processed before model training to obtain a total of 389 groups of sample input data. The input features of the classification type in the sample are first one-hot encoded (one_hot, this step uses the OneHotEncoder() class), and then the continuous input variables are normalized (this step uses the MinMaxScaler() class), and then the data is divided into training and test sets (this step uses the sklearn.model_selection.train_test_split() function, where test_size=0.2), and the data is divided into input_train, input_test, output_train, and output_test.

[0121] Then, the model is trained using the training data (input_train, output_train). The Boosting ensemble model is divided into three basic learners, with a total of 500 iterations K. The basic learners of each iteration are: Super Parent Independent Dependence (SPODE), Logistic Regression, Decision Tree, and Super Parent Independent Dependence.

[0122] For example, in the first round of iteration, the normalized data is input into the Super Parent Independent Dependence (SPODE) model to obtain the first output output_1, and the error between output_1 and the actual output_train is solved. For the samples with incorrect prediction, D t+1 (i) = D t(i)exp(-α t y i h t (x i )) to increase the sample weight so that it is valued in the next learner, whose weight is calculated using the formula To make predictions, when et≤1 / 2, at≥0, as et decreases, at gradually increases. Therefore, the base learner with a lower classification error rate has a greater weight.

[0123] Then, the data is weighted sampled according to the weights calculated in the previous round, so that the samples predicted incorrectly in the previous round are valued in the logistic regression learner. The steps are repeated until the iteration stops. This can accurately improve the public opinion prediction results, thereby providing guidance for urban traffic operation management and taking proactive measures.

[0124] Example 2

[0125] A signal intersection public opinion prediction system based on an integrated algorithm prediction model, comprising:

[0126] Data collection module: obtains characteristic data of all signalized intersections and public opinion information related to signalized intersections;

[0127] Data processing module: pre-process the characteristic data and public opinion information of the signalized intersection, and establish a signalized intersection public opinion data set based on the pre-processed characteristic data and public opinion information of the signalized intersection;

[0128] The signalized intersection public opinion data set includes several sample subsets, each sample subset takes a public opinion information as a sample, and collects characteristic data of relevant signalized intersections as sample features of the sample;

[0129] Sample weighted prediction module: Substitute the sample features in the sample subset into the integrated algorithm prediction model to perform sample weighted prediction in an iterative manner, and wait for the best prediction sample to be obtained after the iteration is completed.

[0130] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. According to the technical essence of the present invention, within the spirit and principles of the present invention, any simple modification, equivalent replacement and improvement made to the above embodiment still falls within the protection scope of the technical solution of the present invention.

Claims

1. A signal intersection public opinion prediction based on an integrated algorithm prediction model, characterized in that: The method specifically comprises the following steps: S1: Obtain the characteristic data of all signalized intersections and public opinion information related to signalized intersections; S2: preprocessing the characteristic data and public opinion information of the signalized intersection, and establishing a signalized intersection public opinion data set based on the preprocessed characteristic data and public opinion information of the signalized intersection; The signalized intersection public opinion data set includes several sample subsets, each sample subset takes a public opinion information as a sample, and collects characteristic data of relevant signalized intersections as sample features of the sample; S3: Substitute the sample features in the sample subset into the integrated algorithm prediction model to perform sample weighted prediction in an iterative manner, and wait for the best prediction sample to be obtained after the iteration is completed.

2. The signal intersection public opinion prediction based on the integrated algorithm prediction model according to claim 1 is characterized in that: In S3, the integrated algorithm prediction model includes a first base learner, a second base learner, and a third base learner; the first base learner, the second base learner, and the third base learner are all used to perform sample prediction on the current sample features according to the weights of the current sample features.

3. The signal intersection public opinion prediction based on the integrated algorithm prediction model according to claim 2 is characterized in that: In S3, in each iteration process, the sample features in the sample subset are sequentially predicted by the first base learner, the second base learner, and the third base learner to obtain sample prediction results, and the predicted samples in the sample prediction results obtained by the third base learner are used as the best prediction samples.

4. The signal intersection public opinion prediction based on the integrated algorithm prediction model according to claim 2 is characterized in that: In S3, the sample weighted prediction process is specifically as follows: In the process of iterating the first base learner, the second base learner, and the third base learner, the weight of the current sample feature in the current base learner is obtained, so that the current base learner performs sample prediction on the sample feature with the weight of the current sample feature, thereby obtaining the sample prediction result of the current base learner, and then the weight of the sample feature in the next base learner is weighted according to the sample prediction result of the current base learner, so that the next base learner performs sample prediction on the sample feature with the weight of the weighted sample feature, and so on.

5. The signal intersection public opinion prediction based on the integrated algorithm prediction model according to claim 4 is characterized in that: The sample prediction result includes the prediction sample. The process of weighting the weight of the sample feature in the next base learner according to the sample prediction result of the current base learner is specifically as follows: The predicted sample obtained by the current base learner is compared with the corresponding sample, and whether the predicted sample is an incorrectly predicted sample is determined based on the comparison result. If so, the weight of the sample feature in the current base learner is increased to obtain the increased weight, and the weight of the sample feature in the next base learner is updated based on the increased weight.

6. The signal intersection public opinion prediction based on the integrated algorithm prediction model according to claim 5 is characterized in that: The first base learner, the second base learner, and the third base learner all have base learner weights. The process of obtaining the weight of the current sample feature in the current base learner is specifically as follows: The weights of the sample features in the current base learner are updated according to the base learner weights of the current base learner itself and the weights increased according to the weights of the sample features in the previous base learner, so as to obtain the weights of the sample features in the current base learner.

7. The signal intersection public opinion prediction based on the integrated algorithm prediction model according to claim 1 is characterized in that: In S2, the process of preprocessing the characteristic data and public opinion information of the signalized intersection includes the following steps: SA1: Perform data cleaning and data desensitization on the characteristic data and public opinion information of signalized intersections, and remove false public opinion information based on personal information and intersection information in the public opinion information; SA2: Determine the public opinion category of each public opinion information and mark it according to the public opinion category; SA3: Determine the intersection feature data contained in the intersection information in each public opinion information based on the intersection features.

8. The signal intersection public opinion prediction based on the integrated algorithm prediction model according to claim 7 is characterized in that: In S2, each sample subset takes a piece of public opinion information as a sample, marks the public opinion category of the sample through the public opinion information, and uses the intersection feature data contained in the intersection information in the public opinion information as the sample feature of the sample.

9. The signal intersection public opinion prediction based on the integrated algorithm prediction model according to claim 2 is characterized in that: The first base learner uses a super-parent dependency estimation model as a base learner, the second base learner uses a logistic regression model as a base learner, and the third base learner uses a decision tree model as a base learner.

10. A signal intersection public opinion prediction system based on an integrated algorithm prediction model, characterized in that: include: Data collection module: obtains characteristic data of all signalized intersections and public opinion information related to signalized intersections; Data processing module: pre-process the characteristic data and public opinion information of the signalized intersection, and establish a signalized intersection public opinion data set based on the pre-processed characteristic data and public opinion information of the signalized intersection; The signalized intersection public opinion data set includes several sample subsets, each sample subset takes a public opinion information as a sample, and collects characteristic data of relevant signalized intersections as sample features of the sample; Sample weighted prediction module: Substitute the sample features in the sample subset into the integrated algorithm prediction model to perform sample weighted prediction in an iterative manner, and wait for the best prediction sample to be obtained after the iteration is completed.

Images