CN117132004A - Public place people stream density prediction method, system and equipment based on neural network - Google Patents

Public place people stream density prediction method, system and equipment based on neural network Download PDF

Info

Publication number
CN117132004A
CN117132004A CN202311401913.XA CN202311401913A CN117132004A CN 117132004 A CN117132004 A CN 117132004A CN 202311401913 A CN202311401913 A CN 202311401913A CN 117132004 A CN117132004 A CN 117132004A
Authority
CN
China
Prior art keywords
data
public place
neural network
comment
people stream
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202311401913.XA
Other languages
Chinese (zh)
Other versions
CN117132004B (en
Inventor
白登辉
王家良
曾丽竹
付韵潮
邱壮
刘艳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sichuan Architectural Design And Research Institute Co ltd
Original Assignee
Sichuan Architectural Design And Research Institute Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sichuan Architectural Design And Research Institute Co ltd filed Critical Sichuan Architectural Design And Research Institute Co ltd
Priority to CN202311401913.XA priority Critical patent/CN117132004B/en
Publication of CN117132004A publication Critical patent/CN117132004A/en
Application granted granted Critical
Publication of CN117132004B publication Critical patent/CN117132004B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/251—Fusion techniques of input or preprocessed data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services
    • G06Q50/265—Personal security, identity or safety
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Tourism & Hospitality (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Software Systems (AREA)
  • Marketing (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • General Business, Economics & Management (AREA)
  • Development Economics (AREA)
  • Computer Security & Cryptography (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Quality & Reliability (AREA)
  • Primary Health Care (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The application discloses a public place people stream density prediction method, a public place people stream density prediction system and public place people stream density prediction equipment based on a neural network, which comprise the following steps: capturing comment data of a certain public place in a preset time period by a plurality of mainstream comment platforms through the Internet, and carrying out emotion analysis on the comment data by adopting an emotion analysis method to obtain emotion tendency data; basic data of a certain public place in a preset time period are acquired: historical people stream density data, historical weather data, historical air temperature data and historical holiday data; processing and fusing the basic data and the emotion tendency data; dividing the fusion data into a training set and a testing set; constructing an LSTM-based neural network, training the LSTM-based neural network based on a training set, and iteratively updating the LSTM-based neural network by adopting an improved gradient descent algorithm in the training process; and (3) carrying out public place people stream density prediction on the data to be predicted by adopting an optimal model. The application improves the prediction accuracy.

Description

Public place people stream density prediction method, system and equipment based on neural network
Technical Field
The application relates to the technical field of data processing and data prediction, in particular to a public place people stream density prediction method, system and equipment based on a neural network.
Background
With the development of socioeconomic performance, population density has also increased at a high rate. In public places with dense people flow, serious crowding is often caused by excessive concentration of people, so that not only is the mood of people in the process of going out influenced, but also great potential safety hazards are brought. Only if people flow density information in public places is effectively mastered, decision basis can be provided for related departments, so that corresponding measures can be rapidly and accurately made.
For people flow density prediction, the traditional method usually adopts a manual observation means, and the method can not meet the rapid response of crowding of people. Along with development of machine learning, people apply various machine learning models to people's stream density prediction, but traditional machine learning models are difficult to fit deep rules of data, accuracy is not high, and for some novel machine learning methods including neural networks, the novel machine learning methods are good at finding deep rules in data, and if original data is insufficient in dimensionality and single in information, the advantages of the neural networks cannot be exerted.
However, people have a positive or negative evaluation of a public place, which has a significant impact on the density of the future people stream. The existing prediction method does not consider the influence of evaluation on the prediction of the people stream density in public places, which further limits the accuracy of the prediction of the people stream density.
In addition, for the prediction of people stream density data, errors are unavoidable, but the consequences of more predictions and less predictions are different, if the predictions are less, but the actual people stream is more, the relevant departments may not be able to make relevant decisions in advance, resulting in congestion accidents.
In view of this, the present application has been made.
Disclosure of Invention
The application aims to solve the technical problems that the existing public place people stream density prediction method is low in prediction accuracy and the prediction data is smaller than a true value. The application aims to provide a public place people stream density prediction method, a system and equipment based on a neural network, which are used for taking evaluation data of the public place into consideration, integrating influence factors such as emotion tendency, historical people stream density data, weather, air temperature, holidays and the like of people with multi-dimension quantification on the public place, training the neural network which is good at time sequence data processing by using the data, improving the prediction accuracy of the neural network, and making the prediction data deviate to be larger than a true value under the condition of ensuring the accuracy, thereby being more beneficial to guiding related departments to make related decisions in advance and avoiding crowded accidents.
The application is realized by the following technical scheme:
in a first aspect, the present application provides a neural network-based public place people stream density prediction method, the method comprising:
grabbing a plurality of main stream comment platforms through Internet to obtain the comment platform within a preset time periodEach date in (a) carrying out emotion analysis on comment data in a certain public place by adopting an emotion analysis method to obtain emotion tendency data;
acquiring basic data of a certain public place in a preset time period, wherein the basic data comprise historical people stream density data, historical weather data, historical air temperature data and historical holiday data;
processing and fusing the basic data and the emotion tendency data to obtain fused data; dividing the fusion data into a training set and a testing set for training and evaluating a subsequent network model;
constructing an LSTM-based neural network, training the LSTM-based neural network based on a training set, and iteratively updating the LSTM-based neural network by adopting an improved gradient descent algorithm in the training process; inputting the test set into a trained LSTM-based neural network, and selecting a model with highest accuracy as an optimal model;
and (3) carrying out public place people stream density prediction on the data to be predicted by adopting an optimal model.
Further, the comment data includes overall comment data for a certain public place and merchant comment data contained for a certain public place.
Further, performing emotion analysis on the comment data by adopting an emotion analysis method to obtain emotion tendency data, wherein the emotion tendency data comprises the following steps:
carrying out emotion analysis on each comment data by adopting an emotion analysis method to obtain a confidence level that the emotion tendency of each comment is forward: date for a public placeIs the first of (2)iConfidence of the overall comment is +.>The method comprises the steps of carrying out a first treatment on the surface of the Date +.>Comprised merchantsuIs the first of (2)jConfidence of the comment is->；
Determining the date of a public place according to the confidence that the emotion tendency of each comment is positiveEmotional general tendency->For (I)>Wherein->Emotion tendency for overall comment data, mean +.>，LRepresenting overall comment data->Is the total number of bars; />Mean value of emotion tendencies of contained merchant comment data，MRepresenting the contained merchant comment data +.>Is the total number of bars; />For the emotional tendency weight of the whole comment data of a certain public place, the weight is->To weight emotional tendency of merchant comment data contained in a public place,and->；
With general trend of emotionEmotional tendency of overall comment data->Emotional tendency of merchant comment data contained +.>Median of overall comment confidence +.>Standard deviation of overall comment confidence +.>Median of merchant comment confidence level included +.>And the standard deviation of the included merchant assessment confidence level +.>Form emotion trend data.
The above technical proposal is thatRepresenting emotional tendency to the whole specified public place should be taken as a main factor, and +.>Representing emotional tendency of merchants in the water bath box, and playing an auxiliary supplementing role. Furthermore, calculate +.>And (3) withThe median, standard deviation of (2) is denoted +.>，/>，/>，/>. People have important influence on the density of the future people stream for emotion evaluation of a public place, and the more positive people evaluate the public place, the greater the probability of the future people stream. And the emotion evaluation has hysteresis, namely the current evaluation data can relatively influence the tomorrow's people stream rather than the current people stream, so that the emotion tendency data is adopted, and the prediction of the future people stream is facilitated. In contrast, the data of the day such as weather and air temperature (because accurate data of the weather and air temperature in the open day cannot be obtained during prediction) has low contribution to the prediction of the people flow in the open day and low relative emotion evaluation. The specific differences are shown in Table 1.
Further, processing and fusing the basic data and the emotion tendencies data to obtain fused data, including:
respectively and independently carrying out normalization processing on emotion tendency data and basic data in a preset time period, and unifying various normalized data to the same dimension to obtain normalized data;
feature fusion is carried out on the normalized data, and the normalized data are spliced front and back to obtain a fused feature vector; the single feature vector after fusion is expressed as:wherein, the method comprises the steps of, wherein,for the date of a certain public place after normalization +.>Is to (1) emotional general tendency, <' > is to (i)>For the date of a certain public place after normalization +.>Emotional tendency of overall comment data, +.>For the date of a certain public place after normalization +.>Median of overall comment confidence, +.>For the date of a certain public place after normalization +.>Standard deviation of overall comment confidence, +.>For the date of a certain public place after normalization +.>Emotional tendency of merchant comment data contained, < +.>For the date of a certain public place after normalization +.>The median of the confidence level of the merchant comments contained,/>For the date of a certain public place after normalization +.>Standard deviation of included merchant evaluation confidence +.>For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical weather data,/->For the date of a certain public place after normalization +.>Historical air temperature data,/->For the date of a certain public place after normalization +.>Historical holiday data.
The corresponding output label is。/>Sequentially take 4 ton-1, obtaining n-4 eigenvectors. And divide it into a training set and a test set.
Further, the LSTM-based neural network comprises an input layer, a hidden layer and an output layer; the number of LSTM neurons of the input layer is the same as the dimension of the input data of the single feature vector after fusion; the number of LSTM neurons of the hidden layer is 50; the LSTM neurons of the output layer are 1 in number and are used for outputting predicted people stream density data; the weights of the connection among the layers of the neural network are initialized through Gaussian distribution, the mean value of the Gaussian distribution is 0, the variance is 1, and the offset is initialized to 0.
Further, in the training process, an improved gradient descent algorithm is adopted to update the LSTM-based neural network in an iterative manner, and the formula is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Indicate->Parameter set of neural network at time of iteration, +.>Indicate->Parameter set of neural network at time of iteration, +.>Representing the rate of learning over the network->Representing a loss function; />For the weighting factor of the loss when the predicted value is greater than the true value,for the weighting factor lost when the predicted value is smaller than the true value, +.>；/>As predicted values for LSTM based neural networks,is a true value.
According to the technical scheme, in order to make the predicted data deviate to be larger than the true value under the condition of ensuring accuracy, relevant departments are guided to make relevant decisions in advance, and congestion accidents are avoided; thus, when training the network, model predictive values of LSTM-based neural networksLess than true value->The loss value generated is greater than the model predictive value of LSTM-based neural network>Greater than the true value->The resulting loss value, i.e. the former, gives the model a greater penalty, thus achieving the effect that the model is more biased at the time of prediction than the true value, rather than being less. The loss on two sides of the self-defined loss function is different, if the input data and the output are not strongly correlated, the vibration of the model output is easy to be caused, and the method adds various factors which are strongly correlated with the future people flow and include emotion tendencies, so that the vibration of the model can be reduced, and the predicted data is biased to be larger than a true value under the condition of ensuring the accuracy of the model.
In a second aspect, the present application further provides a public place people stream density prediction system based on a neural network, where the system uses the public place people stream density prediction method based on the neural network; the system comprises:
the emotion trend data acquisition unit is used for capturing comment data of a certain public place in a preset time period through a plurality of mainstream comment platforms, and performing emotion analysis on the comment data by adopting an emotion analysis method to obtain emotion trend data;
the basic data acquisition unit is used for acquiring basic data of a certain public place in a preset time period, wherein the basic data comprise historical people stream density data, historical weather data, historical air temperature data and historical holiday data;
the data fusion and division unit is used for processing and fusing the basic data and the emotion tendency data to obtain fused data; dividing the fusion data into a training set and a testing set for training and evaluating a subsequent network model;
the model construction and training test unit is used for constructing an LSTM-based neural network, training the LSTM-based neural network based on a training set, and iteratively updating the LSTM-based neural network by adopting an improved gradient descent algorithm in the training process; inputting the test set into a trained LSTM-based neural network, and selecting a model with highest accuracy as an optimal model;
and the people stream density prediction unit is used for predicting the people stream density of the public place by adopting an optimal model.
Further, the comment data includes overall comment data for a certain public place and merchant comment data contained for a certain public place.
Further, in the training process, an improved gradient descent algorithm is adopted to update the LSTM-based neural network in an iterative manner, and the formula is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Indicate->Parameter set of neural network at time of iteration, +.>Indicate->Parameter set of neural network at time of iteration, +.>Representing the rate of learning over the network->Representing a loss function; />For the weighting factor of the loss when the predicted value is greater than the true value,for the weighting factor lost when the predicted value is smaller than the true value, +.>；/>As predicted values for LSTM based neural networks,is a true value.
In a third aspect, the present application further provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the neural network-based public place people stream density prediction method described above when executing the computer program.
Compared with the prior art, the application has the following advantages and beneficial effects:
according to the public place people stream density prediction method, system and equipment based on the neural network, evaluation data of the public place are considered, influence factors such as emotion tendency, historical people stream density data, weather, air temperature, holidays and the like of people with multi-dimension quantification are fused, the neural network good at time sequence data processing is trained by utilizing the data, and the prediction accuracy of the neural network is improved; and the custom loss function is adopted, so that the predicted data is biased to be larger than the true value under the condition of ensuring the accuracy, and the method is more beneficial to guiding related departments to make related decisions in advance, and congestion accidents are avoided.
Drawings
The accompanying drawings, which are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings:
FIG. 1 is a flow chart of a public place people stream density prediction method based on a neural network;
fig. 2 is a block diagram of a public place people stream density prediction system based on a neural network.
Detailed Description
For the purpose of making apparent the objects, technical solutions and advantages of the present application, the present application will be further described in detail with reference to the following examples and the accompanying drawings, wherein the exemplary embodiments of the present application and the descriptions thereof are for illustrating the present application only and are not to be construed as limiting the present application.
The current public place people stream density prediction method has the problems that the prediction accuracy is not high and the prediction data is smaller than the true value. The application designs a public place people stream density prediction method, a public place people stream density prediction system and public place people stream density prediction equipment based on a neural network, considers evaluation data of the public place, fuses influence factors such as emotion tendency, historical people stream density data, weather, air temperature, holidays and the like of people in multi-dimension quantification on the public place, trains the neural network good at time sequence data processing by utilizing the data, and improves the prediction accuracy of the neural network; and the custom loss function is adopted, so that the predicted data is biased to be larger than the true value under the condition of ensuring the accuracy, and the method is more beneficial to guiding related departments to make related decisions in advance, and congestion accidents are avoided.
Example 1
As shown in fig. 1, the public place people stream density prediction method based on the neural network of the application comprises the following steps:
grabbing a plurality of main stream comment platforms through Internet to obtain the comment platform within a preset time periodEach date in (a) carrying out emotion analysis on comment data in a certain public place by adopting an emotion analysis method to obtain emotion tendency data;
acquiring basic data of a certain public place in a preset time period, wherein the basic data comprise historical people stream density data, historical weather data, historical air temperature data and historical holiday data;
processing and fusing the basic data and the emotion tendency data to obtain fused data; dividing the fusion data into a training set and a testing set for training and evaluating a subsequent network model;
constructing an LSTM-based neural network, training the LSTM-based neural network based on a training set, and iteratively updating the LSTM-based neural network by adopting an improved gradient descent algorithm in the training process; inputting the test set into a trained LSTM-based neural network, and selecting a model with highest accuracy as an optimal model;
and (3) carrying out public place people stream density prediction on the data to be predicted by adopting an optimal model.
In specific implementation, the method comprises the following steps:
step 1, aiming at the Wandasi square, in a preset time periodAnd grabbing comment data of a public comment and beauty group platform through the Internet by adopting locomotive collector software on each of 300 days, wherein the comment data comprise comment data for the whole of the Wandan square and comment data of 284 merchants contained in the Wandan square.
Step 2, carrying out emotion analysis on each comment data by adopting a Senta emotion analysis system to obtain a confidence level that emotion tendency of each comment is forward: date for specified public placeIs the first of (2)iConfidence of the overall comment isThe method comprises the steps of carrying out a first treatment on the surface of the Day for specified public placeStage->Comprised merchantsuIs the first of (2)jConfidence of the comment is->The method comprises the steps of carrying out a first treatment on the surface of the The Senta emotion analysis system is an existing open-source emotion analysis system;
step 3, determining the date of the appointed public field according to the confidence that the emotion tendency of each comment is forwardEmotional general tendency->The method comprises the following steps: />Wherein->Emotion tendency of overall comment data, mean value，LRepresenting overall comment data->Is the total number of bars; />Mean +.>，MRepresenting the contained merchant comment data +.>Is the total number of bars; furthermore, calculate +.>And->The median, standard deviation of (2) is denoted +.>，/>，/>，/>；
With general trend of emotionEmotional tendency of overall comment data->Emotional tendency of merchant comment data contained +.>Median of overall comment confidence +.>Standard deviation of overall comment confidence +.>Median of merchant comment confidence level included +.>And the standard deviation of the included merchant assessment confidence level +.>Form emotion trend data.
The above technical proposal is thatRepresenting emotional tendency to the whole specified public place should be taken as a main factor, and +.>Representing emotional tendency of merchants in the water bath box, and playing an auxiliary supplementing role. Furthermore, calculate +.>And (3) withThe median, standard deviation of (2) is denoted +.>，/>，/>，/>. People have important influence on the density of the future people stream for emotion evaluation of a public place, and the more positive people evaluate the public place, the greater the probability of the future people stream. And the emotion evaluation has hysteresis, namely the current evaluation data can relatively influence the tomorrow's people stream rather than the current people stream, so that the emotion tendency data is adopted, and the prediction of the future people stream is facilitated. In contrast, the data of the day such as weather and air temperature (because accurate data of the weather and air temperature in the open day cannot be obtained during prediction) has low contribution to the prediction of the people flow in the open day and low relative emotion evaluation. The specific differences are shown in Table 1.
Step 4, obtaining dateHistorical people stream density data of specified public place, recorded as +.>The unit is a person.
Step 5, acquiring dateWeather data of->Wherein, the positive weather such as clear weather is marked as 1, the neutral weather such as cloudy weather and cloudy weather is marked as 0.5, and the negative weather such as raining, snowing and fogging is marked as 0.
Step 6, obtaining dateTemperature data of (2) is recorded as->The units are degrees celsius.
Step 7, obtaining dateHoliday data->Wherein the section dummy diary is 1 and the working diary is 0.
Step 8, for the steps 3 to 7, in the date sectiondEach type of data obtained is subjected to normalization pretreatment independently, and various normalized data are unified to the same dimension, and the normalization formula is as follows:（/>is the value before data normalization, +.>Is the normalized value of the data,/>、/>Representing the minimum value and the maximum value in the original data, respectively). After normalization, it is marked->，/>，/>，/>，/>，/>，/>，/>，/>，/>，/>The method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>For normalized date of appointed public placedIs to (1) emotional general tendency, <' > is to (i)>For normalized date of appointed public placedEmotional tendency of overall comment data, +.>For normalized date of appointed public placedMedian of overall comment confidence, +.>For normalized date of appointed public placedStandard deviation of overall comment confidence, +.>For normalized date of appointed public placedEmotional tendency of merchant comment data contained, < +.>For normalized date of appointed public placedThe median of the confidence level of the merchant comments contained,/>For normalized date of appointed public placedStandard deviation of included merchant evaluation confidence +.>For normalized date of appointed public placedHistorical people stream density data,/->For normalized date of appointed public placedHistorical weather data,/->For normalized date of appointed public placedHistorical air temperature data,/->For normalized date of appointed public placedHistorical holiday data.
And 9, fusing the basic data and the emotion tendency data, unifying the normalized data to the same dimension, and splicing the normalized data front and back to obtain the fused data, namely forming the feature vector (comprising the input data and the output label) for training the LSTM-based neural network.
Specifically, the single feature vector after fusion, namely the input data is:wherein, the method comprises the steps of, wherein,for the date of the normalized designated public place +.>Is to (1) emotional general tendency, <' > is to (i)>For the date of the normalized designated public place +.>Emotional tendency of overall comment data, +.>For the date of the normalized designated public place +.>Median of overall comment confidence, +.>For the date of the normalized designated public place +.>The standard deviation of the overall comment confidence,for the date of the normalized designated public place +.>Emotional tendency of merchant comment data contained, < +.>For the date of the normalized designated public place +.>The median of the confidence level of the merchant comments contained,/>For the date of the normalized designated public place +.>Standard deviation of included merchant evaluation confidence +.>For the date of the normalized designated public place +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of the normalized designated public place +.>Historical weather data,/->For the date of the normalized designated public place +.>Historical air temperature data,/->For the date of the normalized designated public place +.>Historical holiday data.
The corresponding output label is。/>And taking 4 to 299 in sequence to obtain 296 feature vectors. And is divided into a training set and a testing set, wherein the training set is 208, and the testing set is 88.
Step 10, constructing an artificial neural network based on Long Short-Term Memory (LSTM), wherein the LSTM neuron structure is good at time sequence data processing. The LSTM-based neural network comprises an input layer, a hidden layer and an output layer; the number of LSTM neurons of the input layer is the same as the dimension of the input data of the single feature vector after fusion; the number of LSTM neurons of the hidden layer is 50; the LSTM neurons of the output layer are 1 in number and are used for outputting predicted people stream density data; the weights of the connection among the layers of the neural network are initialized through Gaussian distribution, the mean value of the Gaussian distribution is 0, the variance is 1, and the offset is initialized to 0.
And 11, inputting the training set obtained in the step 9 into the LSTM-based neural network constructed in the step 10 for training, wherein the batch size is 10, and the iteration times are 100, 200 and 300 respectively. In the training process, an improved gradient descent algorithm is adopted to carry out iterative updating on the LSTM-based neural network, and the formula is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Representation ofFirst, theParameter set of neural network at time of iteration, +.>Indicate->Parameter set of neural network at time of iteration, +.>Representing the rate of learning over the network->Representing a loss function; />For the weighting factor of losses when the predicted value is greater than the true value, +.>For the weighting factor lost when the predicted value is smaller than the true value, +.>；/>Is a predicted value of LSTM based neural network, < >>Is a true value.
Particularly, in order to bias the predicted data to be larger than the true value under the condition of ensuring accuracy, the related departments are guided to make related decisions in advance, so that congestion accidents are avoided; the application self-defines the loss functionTo solve the above problems. Thus, when model predictive value of LSTM based neural network is +.>Less than true value->The loss value generated is greater than the model predictive value of LSTM-based neural network>Greater than the true value->The resulting loss value, i.e. the former, gives the model a greater penalty, thus achieving the effect that the model is more biased at the time of prediction than the true value, rather than being less. The loss on two sides of the self-defined loss function is different, if the input data and the output are not strongly correlated, the vibration of the model output is easy to be caused, and the method adds various factors which are strongly correlated with the future people flow and include emotion tendencies, so that the vibration of the model can be reduced, and the predicted data is biased to be larger than a true value under the condition of ensuring the accuracy of the model.
Furthermore, to improve training efficiency, pairs are iteratedAttenuation is carried out, and the attenuation formula is->. After the iterative training is completed, inputting the input data of the test set into a trained LSTM-based neural network to obtain a predicted output label +.>. The average absolute percentage error is used for measuring the accuracy of the model, and the expression is as followsWhereinKRepresenting the number of test sets. And under different iteration times, selecting the model with the highest accuracy as an optimal model for future prediction tasks.
The performance gap of the method of the application is compared with that of other existing methods. As shown in tables 1 and 2. As can be seen from table 1, compared with other methods, the accuracy index of the method is highest; when the training data does not contain weather, air temperature and holiday information, the accuracy is second; when the training data does not contain emotion tendency information, the accuracy rank is third; other prior art methods are arranged in sequence. It can be derived that the application of LSTM-based neural network is superior to the common neural network model and the traditional machine learning model; in training data, the model performance can be improved by adding emotion tendency information and weather, air temperature and holiday information, but the model performance can be improved by adding the emotion tendency information.
As can be seen from Table 2, the method of the present application using the custom loss function can bias the predicted data to be larger than the true value while ensuring accuracy; meanwhile, it can be obtained that if only single historical people flow density data is adopted, the model trained by the self-defined loss function is low in accuracy (namely, vibration of the model is easy to cause), and multiple methods are needed to be used simultaneously (namely, the method) so that the accuracy can be ensured and the predicted data is biased to be larger than a true value.
TABLE 1 Performance comparison Table of the inventive method with different input information and different models
Index (I) Support vector machine Fully connected neural network The method of the application only comprises the historical people stream density data The method of the application does not contain weather, air temperature and holiday information The method of the application does not contain emotion trend information The method of the application
MAPE 16.598% 11.307% 9.849% 6.079% 7.338% 3.384%
TABLE 2 Performance comparison Table of the inventive method with different loss functions and different input information
Index (I) The method of the application only comprises the historical people stream density number According to the conventional loss function The method only contains the historical people stream density data, and adopts the method Custom loss function for inventive method The method of the application adopts the traditional method Loss function The application is that Method
MAPE 8.398% 9.849% 3.598% 3.384%
The number of times the predicted value is greater than the true value (total 90) 41 52 42 56
Example 2
As shown in fig. 2, the present embodiment differs from embodiment 1 in that the present embodiment provides a neural network-based public place people stream density prediction system using the above-described neural network-based public place people stream density prediction method; the system comprises:
the emotion trend data acquisition unit is used for capturing comment data of a certain public place in a preset time period through a plurality of mainstream comment platforms, and performing emotion analysis on the comment data by adopting an emotion analysis method to obtain emotion trend data;
the basic data acquisition unit is used for acquiring basic data of a certain public place in a preset time period, wherein the basic data comprise historical people stream density data, historical weather data, historical air temperature data and historical holiday data;
the data fusion and division unit is used for processing and fusing the basic data and the emotion tendency data to obtain fused data; dividing the fusion data into a training set and a testing set for training and evaluating a subsequent network model;
the model construction and training test unit is used for constructing an LSTM-based neural network, training the LSTM-based neural network based on a training set, and iteratively updating the LSTM-based neural network by adopting an improved gradient descent algorithm in the training process; inputting the test set into a trained LSTM-based neural network, and selecting a model with highest accuracy as an optimal model;
and the people stream density prediction unit is used for predicting the people stream density of the public place by adopting an optimal model.
As a further implementation, the comment data includes overall comment data for a certain public place and merchant comment data contained for a certain public place.
As a further implementation, the improved gradient descent algorithm is adopted in the training process to iteratively update the LSTM-based neural network, and the formula is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Represent the firstParameter set of neural network at time of iteration, +.>Indicate->Parameter set of neural network at time of iteration, +.>Representing the rate of learning over the network->Representing a loss function; />For the weighting factor of losses when the predicted value is greater than the true value, +.>For the weighting factor lost when the predicted value is smaller than the true value, +.>；/>Is a predicted value of LSTM based neural network, < >>Is a true value.
The execution process of each unit is performed according to the flow steps of the public place people stream density prediction method based on the neural network in embodiment 1, and the detailed description is omitted in this embodiment.
Meanwhile, the application also provides computer equipment, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor realizes the public place people stream density prediction method based on the neural network when executing the computer program.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the application, and is not meant to limit the scope of the application, but to limit the application to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the application are intended to be included within the scope of the application.

Claims (10)

1. The public place people stream density prediction method based on the neural network is characterized by comprising the following steps:
capturing comment data of a certain public place in a preset time period by a plurality of mainstream comment platforms through the Internet, and carrying out emotion analysis on the comment data by adopting an emotion analysis method to obtain emotion tendency data;
acquiring basic data of a certain public place in a preset time period, wherein the basic data comprise historical people stream density data, historical weather data, historical air temperature data and historical holiday data;
processing and fusing the basic data and the emotion tendency data to obtain fused data; dividing the fusion data into a training set and a testing set;
constructing an LSTM-based neural network, training the LSTM-based neural network based on a training set, and iteratively updating the LSTM-based neural network by adopting an improved gradient descent algorithm in the training process; inputting the test set into a trained LSTM-based neural network, and selecting a model with highest accuracy as an optimal model;
and carrying out public place people stream density prediction on the data to be predicted by adopting the optimal model.
2. The neural network-based public place people stream density prediction method according to claim 1, wherein the comment data includes overall comment data for a certain public place and merchant comment data contained for a certain public place.
3. The neural network-based public place people stream density prediction method according to claim 2, wherein emotion analysis is performed on the evaluation data by adopting an emotion analysis method to obtain emotion tendency data, and the method comprises the following steps:
carrying out emotion analysis on each comment data by adopting an emotion analysis method to obtain a confidence level that the emotion tendency of each comment is forward: date for a public placeIs the first of (2)iConfidence of the overall comment is +.>The method comprises the steps of carrying out a first treatment on the surface of the Date +.>Comprised merchantsuIs the first of (2)jConfidence of the comment is->；
Determining the date of a public place according to the confidence that the emotion tendency of each comment is positiveIs the overall emotional tendency of (a)For (I)>Wherein->Emotional tendency for overall comment data, +.>，LRepresenting overall comment data->Is the total number of bars; />Emotional tendency for the contained merchant comment data, < +.>，MRepresenting the contained merchant comment data +.>Is the total number of bars; />For the emotional tendency weight of the whole comment data of a certain public place, the weight is->For the emotional tendency weight of merchant comment data contained in a certain public place, the weight is +.>And->；
The emotional tendency data is formed by emotional general tendency, emotional tendency of the whole comment data, emotional tendency of the contained merchant comment data, median of the whole comment confidence, standard deviation of the whole comment confidence, median of the contained merchant comment confidence and standard deviation of the contained merchant comment confidence.
4. The neural network-based public place people stream density prediction method according to claim 1, wherein the processing and fusing the basic data and emotion tendencies data to obtain fused data comprises the following steps:
respectively and independently carrying out normalization processing on emotion tendency data and basic data in a preset time period to obtain normalized data;
feature fusion is carried out on the normalized data, namely the normalized data are spliced front and back to obtain a fused feature vector; the single feature vector after fusion is expressed as:wherein, the method comprises the steps of, wherein,for the date of a certain public place after normalization +.>Is to (1) emotional general tendency, <' > is to (i)>For the date of a certain public place after normalization +.>Emotional tendency of overall comment data, +.>For the date of a certain public place after normalization +.>Median of overall comment confidence, +.>For the date of a certain public place after normalization +.>Standard deviation of overall comment confidence, +.>For the date of a certain public place after normalization +.>Emotional tendency of merchant comment data contained, < +.>For the date of a certain public place after normalization +.>The median of the confidence level of the merchant comments contained,/>For the date of a certain public place after normalization +.>Standard deviation of included merchant evaluation confidence +.>For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical people stream density data,/->For the date of a certain public place after normalization +.>Historical weather data,/->For the date of a certain public place after normalization +.>Historical air temperature data,/->For the date of a certain public place after normalization +.>Historical holiday data.
5. The neural network-based public place people stream density prediction method according to claim 1, wherein the LSTM-based neural network comprises an input layer, a hidden layer, and an output layer; the number of LSTM neurons of the input layer is the same as the dimension of the input data of the single feature vector after fusion; the number of LSTM neurons of the hidden layer is 50; the number of LSTM neurons of the output layer is 1, and the LSTM neurons are used for outputting predicted people stream density data; the weights of the connection among the layers of the neural network are initialized through Gaussian distribution, the mean value of the Gaussian distribution is 0, the variance is 1, and the offset is initialized to 0.
6. The neural network-based public place people stream density prediction method according to claim 1, wherein the neural network based on the LSTM is iteratively updated by adopting an improved gradient descent algorithm in a training process, and the formula is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Represent the firstParameter set of neural network at time of iteration, +.>Indicate->Parameter set of neural network at time of iteration, +.>Representing the rate of learning over the network->Representing a loss function; />For the weighting factor of losses when the predicted value is greater than the true value, +.>For the weighting factor lost when the predicted value is smaller than the true value, +.>；/>Is a predicted value of LSTM based neural network, < >>Is a true value.
7. Public place people stream density prediction system based on neural network, characterized by that, this system includes:
the emotion trend data acquisition unit is used for capturing comment data of a certain public place in a preset time period through a plurality of mainstream comment platforms, and performing emotion analysis on the comment data by adopting an emotion analysis method to obtain emotion trend data;
the system comprises a basic data acquisition unit, a control unit and a control unit, wherein the basic data acquisition unit is used for acquiring basic data of a certain public place in a preset time period, and the basic data comprise historical people stream density data, historical weather data, historical air temperature data and historical holiday data;
the data fusion and division unit is used for processing and fusing the basic data and the emotion tendency data to obtain fused data; dividing the fusion data into a training set and a testing set;
the model construction and training test unit is used for constructing an LSTM-based neural network, training the LSTM-based neural network based on a training set, and iteratively updating the LSTM-based neural network by adopting an improved gradient descent algorithm in the training process; inputting the test set into a trained LSTM-based neural network, and selecting a model with highest accuracy as an optimal model;
and the people stream density prediction unit is used for predicting the people stream density of the public place by adopting the optimal model.
8. The neural network-based public place people stream density prediction system of claim 7, wherein the comment data includes overall comment data for a certain public place and merchant comment data contained for a certain public place.
9. The neural network-based public place people stream density prediction system of claim 7, wherein the improved gradient descent algorithm is used for iteratively updating the LSTM-based neural network during training, the formula being:the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Represent the firstParameter set of neural network at time of iteration, +.>Indicate->Parameter set of neural network at time of iteration, +.>Representing the rate of learning over the network->Representing a loss function; />For the weighting factor of losses when the predicted value is greater than the true value, +.>Is small in predicted valueWeight coefficient lost at true value, +.>；/>Is a predicted value of LSTM based neural network, < >>Is a true value.
10. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the neural network-based public place people stream density prediction method of any one of claims 1 to 6 when the computer program is executed.
CN202311401913.XA 2023-10-27 2023-10-27 Public place people stream density prediction method, system and equipment based on neural network Active CN117132004B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311401913.XA CN117132004B (en) 2023-10-27 2023-10-27 Public place people stream density prediction method, system and equipment based on neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311401913.XA CN117132004B (en) 2023-10-27 2023-10-27 Public place people stream density prediction method, system and equipment based on neural network

Publications (2)

Publication Number Publication Date
CN117132004A true CN117132004A (en) 2023-11-28
CN117132004B CN117132004B (en) 2024-02-09

Family

ID=88860424

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311401913.XA Active CN117132004B (en) 2023-10-27 2023-10-27 Public place people stream density prediction method, system and equipment based on neural network

Country Status (1)

Country Link
CN (1) CN117132004B (en)

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107066565A (en) * 2017-04-01 2017-08-18 上海诺悦智能科技有限公司 A kind of tourist hot spot forecasting system
AU2018100320A4 (en) * 2018-03-15 2018-04-26 Ji, Jiajian Mr A New System for Stock Volatility Prediction by Using Long Short-Term Memory with Sentimental Indicators
CN110598775A (en) * 2019-09-03 2019-12-20 合肥工业大学 Prediction method, system and storage medium based on fuzzy clustering and BP neural network
US20200104369A1 (en) * 2018-09-27 2020-04-02 Apple Inc. Sentiment prediction from textual data
US20200394455A1 (en) * 2019-06-15 2020-12-17 Paul Lee Data analytics engine for dynamic network-based resource-sharing
CN112632377A (en) * 2020-12-21 2021-04-09 西北大学 Recommendation method based on user comment emotion analysis and matrix decomposition
US20220019946A1 (en) * 2020-07-20 2022-01-20 UST Global Inc Systems and methods for generating and updating travel itineraries
CN114443844A (en) * 2022-01-18 2022-05-06 安徽大学 Social network comment text sentiment analysis method and system fusing user sentiment tendency
CN116011447A (en) * 2023-03-28 2023-04-25 杭州实在智能科技有限公司 E-commerce comment analysis method, system and computer readable storage medium
WO2023134083A1 (en) * 2022-01-11 2023-07-20 平安科技（深圳）有限公司 Text-based sentiment classification method and apparatus, and computer device and storage medium

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107066565A (en) * 2017-04-01 2017-08-18 上海诺悦智能科技有限公司 A kind of tourist hot spot forecasting system
AU2018100320A4 (en) * 2018-03-15 2018-04-26 Ji, Jiajian Mr A New System for Stock Volatility Prediction by Using Long Short-Term Memory with Sentimental Indicators
US20200104369A1 (en) * 2018-09-27 2020-04-02 Apple Inc. Sentiment prediction from textual data
US20200394455A1 (en) * 2019-06-15 2020-12-17 Paul Lee Data analytics engine for dynamic network-based resource-sharing
CN110598775A (en) * 2019-09-03 2019-12-20 合肥工业大学 Prediction method, system and storage medium based on fuzzy clustering and BP neural network
US20220019946A1 (en) * 2020-07-20 2022-01-20 UST Global Inc Systems and methods for generating and updating travel itineraries
CN112632377A (en) * 2020-12-21 2021-04-09 西北大学 Recommendation method based on user comment emotion analysis and matrix decomposition
WO2023134083A1 (en) * 2022-01-11 2023-07-20 平安科技（深圳）有限公司 Text-based sentiment classification method and apparatus, and computer device and storage medium
CN114443844A (en) * 2022-01-18 2022-05-06 安徽大学 Social network comment text sentiment analysis method and system fusing user sentiment tendency
CN116011447A (en) * 2023-03-28 2023-04-25 杭州实在智能科技有限公司 E-commerce comment analysis method, system and computer readable storage medium

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
ANA VALDIVIA ET AL.: "Sentiment Analysis in TripAdvisor", 《IEEE INTELLIGENT SYSTEMS》, vol. 32, no. 4, pages 72, XP011659227, DOI: 10.1109/MIS.2017.3121555 *
徐硕: "基于多特征融合的景区客流预测", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》, no. 02, pages 140 - 699 *
熊奕洋: "基于深度学习的城市人流量预测", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》, no. 05, pages 138 - 1180 *
王铃;陶宏才;: "基于LSTM前融合中文情感倾向分类模型的研究", 成都信息工程大学学报, no. 02, pages 14 - 20 *
贾静: "基于多维混合神经网络的城市人群流量预测研究与实现", 《中国优秀硕士学位论文全文数据库 (社会科学Ⅱ辑)》, no. 12, pages 125 - 99 *

Also Published As

Publication number Publication date
CN117132004B (en) 2024-02-09

Similar Documents

Publication Publication Date Title
CN110262233B (en) Optimization method for technological parameters of magnetic control film plating instrument
CN111047085B (en) Hybrid vehicle working condition prediction method based on meta-learning
CN112257341A (en) Customized product performance prediction method based on heterogeneous data difference compensation fusion
CN110119540B (en) Multi-output gradient lifting tree modeling method for survival risk analysis
CN109740160B (en) Task issuing method based on artificial intelligence semantic analysis
CN110110372B (en) Automatic segmentation prediction method for user time sequence behavior
CN110222838B (en) Document sorting method and device, electronic equipment and storage medium
US20230401637A1 (en) Deep learning approach for assessing credit risk
CN113610552A (en) User loss prediction method and device
CN113761388A (en) Recommendation method and device, electronic equipment and storage medium
CN114004153A (en) Penetration depth prediction method based on multi-source data fusion
CN112215412A (en) Dissolved oxygen prediction method and device
CN111126758B (en) Academic team influence propagation prediction method, academic team influence propagation prediction equipment and storage medium
Moroz et al. Hybrid sorting-out algorithm COMBI-GA with evolutionary growth of model complexity
CN116993548A (en) Incremental learning-based education training institution credit assessment method and system for LightGBM-SVM
CN117132004B (en) Public place people stream density prediction method, system and equipment based on neural network
CN111967973B (en) Bank customer data processing method and device
CN113656707A (en) Financing product recommendation method, system, storage medium and equipment
CN106600100B (en) Weighted multi-population particle swarm optimization-based hazard source reason analysis method
CN113392958B (en) Parameter optimization and application method and system of fuzzy neural network FNN
CN110781978A (en) Feature processing method and system for machine learning
CN114648178B (en) Operation and maintenance strategy optimization method of electric energy metering device based on DDPG algorithm
CN109902870A (en) Electric grid investment prediction technique based on AdaBoost regression tree model
CN113435927B (en) User willingness prediction method, device, equipment and storage medium
JP7214672B2 (en) Information processing device, information processing method, and computer program

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
REG Reference to a national code

Ref country code: HK

Ref legal event code: DE

Ref document number: 40104039

Country of ref document: HK