56 results about "Regression tree model" patented technology

The invention discloses a data mining-based drought monitoring method. The method comprises the following steps: 1, data reconstruction is carried out on an MODIS vegetation index product, a land surface temperature product and an evapotranspiration product; 2, according to the vegetation index obtained in the first step and DEM data, downscaling is carried out on a TRMM rainfall product; 3, a vegetation anomaly index, a temperature anomaly index, an evapotranspiration anomaly index and a rainfall anomaly index are extracted again; and 4, a classification and regression tree model is used for building a statistical regression rule and a linear fitting model to obtain a drought monitoring model. Compared with the prior art, the method of the invention comprehensively considers multi-source remote sensing spatial information, such as the rainfall, the evapotranspiration, the vegetation growth state, the land using type, the altitude and other factors, in the case of drought monitoring, spatial data mining is adopted, the drought monitoring model is built, and the drought monitoring precision is improved.

The invention provides a virtualized cloud platform energy consumption measurement method based on tree regression and a virtualized cloud platform energy consumption measurement system based on the tree regression. The method comprises the following steps: information collection, namely, sending acquired information to assigned control nodes by physical machines, wherein the acquired information comprises the resource information of the physical machines, the resource information of virtual machines, and the energy consumption information obtained from meters; parameter training, namely, the control nodes being responsible for implementing linear fitting on a data set in sections by using a tree regression algorithm to obtain resource energy consumption model parameters alpha and gamma in each subset; energy consumption calculation, namely, implementing calculation by the control nodes according to the fitted parameters and the information of the virtual machines to obtain the energy consumption of the virtual machines. The method and the system have the beneficial effects that as linear models are established in sections for the usage rates of various resources by using a method based on the tree regression, the defect that a traditional single linear model is not high enough in precision is overcome; a regression tree model is simple and efficient to calculate, and reflects a higher efficiency of updating real-time energy consumption model parameters.

The invention relates to a network link quality prediction method and device, and a readable storage medium. The network link quality prediction method includes the following steps: acquiring link quality sample data in multiple different experimental scenes, and performing link quality grading on the link quality sample data by using a non-supervision clustering algorithm; calculating and evaluating the link quality sample data by using a random forest classification algorithm to obtain a corresponding link quality grade value, and extracting a training set from the link quality sample data to construct a combined classification model; and constructing a random forest regression tree model, and predicting the link quality grade value at the next moment according to the link quality gradevalue and the output result of the combined classification model. The network link quality prediction method can effectively reduce man-made interference during the link quality prediction process, soas to improve the prediction efficiency and the prediction accuracy.

The invention relates to the technical field of computer vision, medical image processing, key point detection and the like, and provides a method for automatically positioning a head shadow positioning point. The method for automatically positioning a head shadow measurement mark point comprises the following steps: step 1, preparing data of an X-ray head side position film; step 2, acquiring marking points of doctors in the training set, and performing feature selection; step 3, training a first-layer regression tree model, updating positions of all mark points once every time the first-level regression tree model passes through a cascade regression device to enable the mark points to be closer to the mark points, and calculating a difference value between a current shape and a real shape as a residual error; step 4, training a regression device of each stage in the cascade network; and step 5, updating the current shape to be the current shape plus the residual error, and constructing a cascade residual error regression tree until splitting to leaf nodes. The method is mainly applied to the occasion of automatically positioning the head shadow survey mark point.

The invention provides a wind power generation prediction method and product based on a cost-oriented gradient rising regression tree, and the method comprises the steps: obtaining the wind power historical data of a to-be-predicted place, and predicting an error loss function model; solving a negative gradient value of the wind power historical data with respect to the prediction error loss function model as a residual error estimation value; training a gradient rising regression tree model by using the residual estimation value to obtain a cost-oriented gradient rising regression tree model;and predicting the wind power generation amount by using the cost-oriented gradient rising regression tree model. According to the invention, a gradient rising regression tree method is adopted; according to the method, the cost-oriented loss function can be effectively processed; two means of regression tree and gradient lifting are used for bringing the actual cost generated by the prediction error into the model construction and prediction process, and the gradient rising regression tree method is used for executing the optimal point prediction, so that the cost-oriented loss function canbe effectively processed, and the cost difference caused by high-estimation and low-estimation prediction can be reduced.

The invention relates to an oil and gas drilling machine drilling speed prediction and optimization method based on a CART algorithm. The method comprises the steps: 1, acquiring data; 2, carrying outdata preprocessing separately, taking the eight drilling parameters as different characteristic attributes and taking drilling data contained in the eight drilling parameters as input variables X, taking the mechanical drilling speed as an output variable Y, and obtaining an initial data set D1; 3, performing data correlation analysis to obtain training data sets D2 of different opening times; step 4, establishing a regression tree model between the input variable and the mechanical drilling speed in the training data set D2 with different opening times by utilizing a CART algorithm; 5, analyzing each piece of leaf node information of the generated binary tree, wherein the mean value of the leaf nodes is used as a predicted value of the mechanical drilling speed; 6, traversing the node division result of each layer from top to bottom to obtain different drilling parameter recommendation values; and step 7, obtaining optimal judgment of the mechanical drilling speed. The drilling period can be shortened, the drilling cost is reduced, and therefore the development efficiency of oil and gas resources is greatly improved.

The invention discloses a microblog rumor detection algorithm based on gradient boosting trees and a rumor detection feature set. The provided feature set of rumor detection contains 23 features. According to the provided rumor detection algorithm based on the gradient boosting trees, firstly, training samples are constructed according to the features in the feature set, and the training samples are used for training of a microblog rumor detection model; then multiple training is carried out on the training sample set to obtain multiple regression tree models, each regression tree gives a predicted value, and the predicted values of the multiple regression trees are combined for obtaining the final microblog rumor detection model; and in rumor detection, features of a to-be-predicted microblog are extracted according to the feature set, the detection model is used to calculate and derive a predicted value on the to-be-predicted microblog, and according to the predicted value, judging that the to-be-predicted microblog belongs to rumor microblogs or non-rumor microblogs. Compared with existing microblog rumor detection algorithms, the microblog rumor detection algorithm based on thegradient boosting trees and the rumor detection feature set provided by the invention can bring higher rumor detection precision, and especially in an early period of releasing a rumor, detection precision is significantly higher than that of the existing microblog rumor detection algorithms.

The invention discloses a traditional Chinese medicinal material seed distinguishing and grade rapid distinguishing method, which comprises the following steps: randomly selecting every 100 seeds fromradix peucedani and radix saposhnikoviae seeds as a sample, and respectively constructing 18 samples for each type of seeds; analyzing the sample by using a hyperspectral imaging technology to obtainsample information of different types of seeds; furthermore, the ROI area of each sample is preferably selected; the method comprises the following steps: carrying out averaging processing on spectrums of an area to construct characteristic spectrum information of the area, fusing two types of characteristics by adopting a kernel fusion technology, constructing a plurality of classification regression tree models for optimizing kernel fusion parameters, and establishing a classification prediction model in an optimization process, so that the discrimination accuracy of different types of seeds reaches 92%. According to the method, the hyperspectral imaging technology is utilized to obtain the sample image features and the spectral features at the same time, the multi-feature extraction and fusion technology is adopted, the random forest algorithm is combined to establish the discrimination model, and rapid and lossless discrimination of different types of seeds is achieved.

The invention provides an enterprise financial index capital amount prediction method, device and equipment and a storage medium. The method comprises the steps: acquiring transaction data and auxiliary data of a cash center; performing feature extraction on the transaction data and the auxiliary data to obtain feature data; respectively inputting the feature data into a plurality of pre-trained monomer prediction models to obtain corresponding prediction values; adopting a non-dominated sorting genetic algorithm with an elitist strategy to process the prediction value of each single prediction model to obtain a cash inventory use amount prediction result, wherein the monomer prediction model is a ridge regression model, a classification regression tree model, an extreme random regressiontree model, a LightGBM regression model or a convolutional neural network model; the budget result of each index of an enterprise in a future period of time can be accurately predicted, so the optimalallocation of capital budget of each financial index of the enterprise in the future is realized, and the enterprise is helped to improve the management efficiency.

The invention provides a power battery fault diagnosis method and early warning system based on data driving. The method comprises the following steps: 1, collecting the performance parameters of a power battery under various working conditions and various states of the power battery, including the capacity, voltage, internal resistance and power of the power battery; 2, cleaning the acquired data; 3, calculating the state of charge SOC and the state of health SOH of the power battery according to the cleaned data; 4, formulating a fault level according to actual driving experience and automobile safety; 5, making the data obtained in steps 2, 3 and 4 into a data set; 6, putting the training set into a gradient boosting regression tree model, and carrying out iterative training on the training set; and 7, putting the test set into the model, evaluating the accuracy of the model, and adjusting model parameters according to the accuracy. The power battery fault can be accurately predicted, early warning is carried out on the fault, and the safety of the electric vehicle is greatly improved.

The invention relates to the technical field of logging engineering, aims to provide a formation pore pressure prediction method based on machine learning, and solves the problems that an existing prediction method is lower in prediction result accuracy and not ideal in effect. According to the technical scheme, the formation pore pressure prediction method based on machine learning comprises the following prediction steps of a, processing and preparing data, namely collecting the related logging data and the related rock physical property parameters; b, determining a sensitive curve, namely preparing a reference sequence and a comparison sequence of a grey relational degree method, and determining a sensitive logging curve; c, training and testing a model, namely dividing an original data set into a training set and a testing set, and inputting the training set into a gradient boosting regression tree model to obtain an optimal model; and d, predicting the formation pore pressure, namely taking the sensitive logging curve as an input feature vector of the optimal model to predict the reservoir formation pressure. The method has the advantages of better prediction precision, wide prediction range, high reliability and the like.

The invention relates to the technical field of motion behavior recognition, in particular to an animal motion behavior discrimination method and a device based on time series correlation analysis. The method comprises obtaining the known continuous motion state of the animal, the corresponding three-axis acceleration information and the continuous three-axis acceleration information to be discriminated; converting three-axis acceleration information into three-dimensional data in natural coordinate system, forming training data with three-axis acceleration information of known motion state and corresponding three-dimensional data, forming test data with three-axis acceleration information to be discriminated and corresponding three-dimensional data; according to the temporal correlation between the movement behavior states and the iterative regression tree model trained by the training data, judging the movement behavior states of the test data. By combining with the temporality of behavior state, the method makes up for the shortcomings of iterative decision tree in the processing of highly correlated data, and further improves the accuracy of behavior discrimination.

The invention discloses an urban bus emission rate estimation method based on a gradient lifting regression tree. Firstly, according to the measured bus emission data, a Lagrangian interpolation method is used for standardized processing to obtain the emission data per second. Secondly, the VSP (Vehicle Specific Power) is used to characterize the current operating conditions of the bus, and the influence of the previous driving state on the emission is considered to establish a quantitative model of the emission rate. Finally, the gradient lifting regression tree is used to train data and adjus the parameters, the bus emission rate estimation model is obtained. The invention considers the common influence of the current time operation condition and the previous driving state on the currenttime emission rate, The non-parametric method of gradient lifting regression tree model is used to improve the estimation accuracy of bus emission rate, which has practical significance for controlling traffic exhaust emissions and optimizing road environment, and overcomes the complex nonlinear relationship between bus emission rate and various influencing factors.

The invention discloses a binary classification oriented factor screening method based on boosted regression trees. The method comprises the following steps that: (1) searching data, and establishinga target variable-predictive factor dataset; (2) on the basis of the target variable and all factors, utilizing the boosted regression trees to carry out modeling, and calculating and sorting factor importance; (3) carrying out correlation analysis on all factors, analyzing a Pearson correlation matrix, and carrying out screening; (4) on the basis of the target variable and the retained factor, utilizing the boosted regression trees to establish a new model, calculating a predictive deviation, calculating and sorting the factor importance, and removing the factor with the lowest importance until the amount of the retained factors is less than or equal to 2; and (5) comparing the predictive deviation of each boosted regression tree model in the (4), and taking all factors adopted by the boosted regression tree model with the smallest predictive deviation as an optimal factor combination. By use of the method, a quantitative factor selection system is established, results are reliable, and an application field is wide.

The invention relates to a vegetation index prediction method, a system and a device based on a classification and regression tree algorithm. The method comprises the steps of through using vegetationindexes as dependent variables; and constructing a classification and regression tree model by taking the global land data assimilation system basin surface model data set and the elevation data as independent variables, classifying the sample data by utilizing the classification and regression tree model, predicting a vegetation index of a target time period according to a classification result,and obtaining a vegetation index prediction result. Compared with the prior art, the problem of lack of vegetation indexes in the prior art is solved, a user can use the method to realize vegetationindex prediction in any time period, and vegetation index data are perfected.

The present invention relates to a method for predicting and optimizing a penetration rate in oil and gas drilling based on a CART algorithm. The method includes the following steps: Step 1: collecting data; Step 2: performing data preprocessing in spuds, and obtaining an initial data set D1 by taking 8 drilling parameters as different characteristic attributes and drilling data contained in each characteristic attribute as input variables X and the penetration rate as an output variable Y; Step 3: performing correlation analysis on data to obtain a training data set D2 of different spuds; Step 4: establishing a regression tree model between the input variables and the penetration rate in the training data set D2 of different spuds by using the CART algorithm; Step 5: analyzing information of each leaf node of the generated binary tree, wherein an average value of the leaf nodes is used as a predicted value of the penetration rate; Step 6: traversing a node division result of each layer from top to bottom to obtain different recommended values of drilling parameters; and Step 7: performing optimal judgment of the penetration rate. The method provided by the present invention can shorten the drilling cycle and reduce the drilling cost, thereby greatly improving the development efficiency of oil and gas resources.

The invention discloses a power grid investment prediction method based on an AdaBoost regression tree model. The method comprises steps of determining N power grid investment related technical indexes according to requirements, acquiring numerical values of the N power grid investment related technical indexes at M time points and corresponding power grid investments, constructing a power grid investment related technical index vector at each time point; carrying out dimensionless processing; forming a training sample set by the dimensionless power grid investment related technical index vectors and the power grid investments; and training the AdaBoost regression tree by adopting the training sample set, performing K times of iteration to obtain K AdaBoost regression trees as weak learners, selecting the weak learners from the K AdaBoost regression trees to obtain a strong learner, and performing power grid investment prediction by utilizing the strong learners. According to the method, the training process of the AdaBoost regression tree model is improved, the training process is introduced into power grid investment prediction, and the power grid investment prediction accuracy is improved.

The invention discloses a time sequence database-oriented query time prediction method, relates to the technical field of computers, and aims to solve the problem of low query time prediction speed in the prior art. 2, the time series data are written into a CnosDB, the CnosDB uses a CnoSQL query statement to conduct query retrieval on the time series data, and query time is recorded; 3, encoding the query statement into vectorized data; 4, extracting data distribution characteristics of the vectorized data; 5, performing dimension reduction on the data distribution characteristics by using PCA; step 6, using the vectorized data and the dimension-reduced data distribution characteristics as input, using query time as output, and training a gradient lifting regression tree model; and 7, performing query time prediction by using the trained gradient lifting regression tree model. In the aspect of prediction time, the model can give a prediction result within dozens of milliseconds in the experiment, and the response speed is very considerable.

The invention discloses a cruising taxi transport power scale dynamic adjusting method; the method uses operation data of a taxi information management system as basis, fully considers single car daily working trips, a mileage utilization rate, an earning level, a single car average operation duration, a single trip waiting time duration, and an online car hailing business shared ratio, employs adata mining-decision tree method to construct a cruising taxi transport power scale adjusting regression tree model, and proposes a cruising taxi dynamic adjusting mechanism and thresholds and rankingrelations of key indexes, thus solving the non-unified system problems of the cruising taxi transport power scale adjusting indexes, and providing clear quantitative standards for each key index threshold and importance ranking relations between the indexes. Compared with the prior art, the method can scientifically and reasonably rank the importance of each key index relevant to the taxi transport power scale, thus obtaining more accurate taxi transport power scale predicted values, and further providing theory basis and decision supports for adjusting the city cruising taxi transport powerscale.

The invention discloses a road feeling simulation method based on a GMM and a CART regression tree. The method comprises the steps: carrying out the real vehicle testing and data collection, carrying out the preprocessing of test data, carrying out the normalization of test data clustering, dividing a training and testing data set, training and testing a road feeling simulation model based on the GMM and the CART regression tree, and judging whether the obtained road feeling model meets the requirements or not. performing road feel simulation according to the obtained road feel simulation model based on the GMM and the CART regression tree, wherein the input variables of the CART regression tree model are longitudinal vehicle speed, vehicle transverse acceleration, vehicle yaw velocity, vehicle vertical load, steering wheel angle and steering wheel angular velocity, and the output variables of the CART regression tree model are steering wheel torque. Tests prove that the road feel simulation model based on the GMM and the CART regression tree, which is obtained by the method, is relatively high in precision, the modeling process is easy to implement, and the defects in the prior art are overcome to a certain extent.

The invention discloses a TBM tunneling optimization method based on rock slag physical characteristics. The method comprises the steps that firstly, image acquisition and sensor equipment of a system is installed, and TBM field tunneling parameter data and parameter data of geometric characteristics and physical characteristics of rock slag are acquired to serve as a sample set of a model; secondly, a gradient lifting regression tree model optimized by a particle swarm algorithm is established for parameter learning and training feedback, and a TBM tunneling parameter suggestion interval is controlled; thirdly, the TBM net tunneling rate is output, an optimal prediction model is obtained, the working performance of the optimal prediction model is evaluated according to a test set in samples, and optimal tunneling control parameters are provided; and finally, after the optimal tunneling control parameters are compared with related specification requirements, feedback is conducted to a TBM console in time, and TBM tunneling parameters are adjusted. Optimization provided by the invention can be applied to TBM construction, rock slag information is predicted in advance, the tunneling parameters are dynamically adjusted, intelligent prediction of the TBM rock breaking efficiency is achieved, and the method has important significance in safe and efficient construction of tunnels.

The invention relates to the technical field of data classification, and provides a method and a device for learning rate calculation, and a method and a device for classification model calculation. The learning rate calculation method comprises: obtaining empirical risk of a classification model for classifying data, parameters of the empirical risk including learning rate for iteratively calculating the classification model; performing iterative computation on the learning rate based on random walk, and obtaining a learning rate value when the empirical risk is minimum. Because random walksare introduced into the iterative computation of learning rate, the learning rate obtained from the computation can converge the empirical risk which is used as an optimization object to a global optimal solution rather than a local optimal solution. Furthermore, the classification model can obtain a higher precision model when the model performs iterative calculation based on the learning rate, so as to improve accuracy of data classification results. The method for classification model calculation is used to calculate a gradient progressive regression tree model, and the method for learningrate calculation is used when the learning rate of the model is calculated.

The invention relates to a method and a device for predicting dangerous weather events based on a multiple incremental regression tree model. The method comprises the following steps: 1) reading the historical meteorological observation data and taking the meteorological characteristic data and the dangerous weather event record as the sample data set; 2) establish a training data matrix and a verification data matrix according to that sample data set; 3) setting model parameters of the multiple incremental regression tree model; 4) input a training data matrix and a verification data matrix,training that multiple incremental regression tree model to obtain a trained multiple incremental regression tree model; 5) inputting the prediction data matrix to the trained multiple incremental regression tree model to obtain the occurrence probability of the future dangerous weather event. The invention can significantly improve the prediction accuracy rate of the dangerous weather event.

The invention relates to method for forecasting reaction property of a solid fuel during the chemical chain process. The method comprises the steps of (1) collecting data by solid fuel chemical chainexperiment study; (2) organizing the data to obtain a training sample and a test sample; (3) training the training sample by a gradient boosted regression tree model; and (4) forecasting reaction property of the solid fuel during the chemical chain process. A result is forecasted by traversing data combinations, and a corresponding chemical chain working condition parameter is obtained according to demands of different chemical chain technologies; compared with the prior art, the method has the advantages that the reaction properties of various solid fuels during the chemical chain process areforecasted by the gradient boosted regression tree model, the experiment number is substantially reduced, and labor and material are greatly saved; and moreover, the fuel conversion rate during the chemical chain process is favorably, intuitively and quantitatively forecasted, and the method has an important significance to optimization of the chemical chain process.

The invention belongs to the field of energy, and particularly relates to a visualization method for energy load prediction. The visualization method comprises the following steps: forming an energy load prediction model based on a classification and regression tree model and gradient lifting; inputting energy load historical data parameters and real-time parameters the formed energy load prediction model based on the classification and regression tree model and the gradient lifting; and performing visualization processing on a prediction path and a prediction result of the energy load prediction model based on the classification and regression tree model and the gradient lifting. According to the method, the energy load prediction process can be visually displayed.

The invention discloses a multiphase flow virtual metering method based on a gradient boosting regression tree model. The method comprises the following steps: firstly, acquiring sample data of a plurality of multiphase flows including pressure, temperature, differential pressure, liquid quantity, oil quantity and water quantity; using the sample data for training a GBDT gradient boosting regression tree model to obtain a virtual metering model, using the sample data for evaluating the virtual metering model, and finally the pressure, inputting the real-time temperature and the real-time differential pressure measured in real time into the virtual metering model for virtual metering. The method has the advantages of being high in model training speed, small in parameter adjustment, high inprecision, small in error, small in equipment size, low in manufacturing cost, non-radiative and the like, and has high data reference value for production dynamic monitoring, flow guarantee and oilreservoir management of an oil field.

This disclosure includes systems for regression-tree-modified feature vector machine learning models for utilization prediction in time-expiring inventory. An online computing system receives a feature vector for a listing and inputs the feature vector and modified feature vectors into a demand function to generate demand estimates. The system inputs the demand estimates into a likelihood model togenerate a set of request likelihoods, each request likelihood representing a likelihood that the time-expiring inventory will receive a transaction request at each of a set of test price and test times to expiration. The system further trains a regression tree model based on a set of training data comprising each of the request likelihoods from the set and the test price and test time period toexpiration used to generate the demand estimate that was used to generate the request likelihood.

The invention provides a user information processing method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining user information of a target user, with the user information at least comprising user personal information and historical financial behavior information; inputting the user information into a pre-trained user income level prediction model for feature learning, and outputting income level information of the target user; wherein the user income level prediction model is obtained by training a regression tree model by using a user sample set containing user personal information, historical financial behavior information, a credit line of a credit card and historical income information. The technical problem of inaccurate user income level prediction of an existing rule model is solved.

The invention discloses a road feeling simulation method based on K-Medoids and a classification regression tree. The method comprises the steps of carrying out a real vehicle test, collecting data, carrying out the preprocessing of test data, carrying out the clustering of normalized test data, dividing training and testing data sets, training and testing a road feeling simulation model based on the K-Medoids and CART regression tree, judging whether the obtained road feeling model meets the requirements or not, and performing road feeling simulation according to the obtained road feeling simulation model based on the K-Medoids and CART regression tree. The input variables of the CART regression tree model are the vehicle longitudinal speed, the vehicle transverse acceleration, the vehicle yaw velocity, the vehicle vertical load, the steering wheel angle and the steering wheel angular velocity, and the output variable of the CART regression tree model is the steering wheel torque. Tests prove that the obtained road feeling simulation model based on the K-Medoids and CART regression tree is high in precision, the modeling process is easy to implement, and the defects in the prior art are overcome to a certain extent.

The invention discloses a power distribution network voltage reactive power optimization method based on a BART algorithm. The BART algorithm is an integrated predication method, an accumulative regression tree model is decomposed into a plurality of weak regression trees through a non-parametric Bayesian regression method, and an integrated predication system is formed through an integrated method. Each weak regression tree is in charge of a small part in the entire integrated predication system, and the influence, on predication, of a single regression tree module is weakened to improve the predication effect of the entire integrated model. According to the power distribution network voltage reactive power optimization method based on the BART algorithm, based on self learning of historical sample data and under the premise that voltages are in a reasonable controlled range is guaranteed, the actions of adjusting a transformer tap and the actions of switching of a capacitor bank are as less as possible, under the idea state that the voltages on the lower voltage side of a transformer and the reactive power are small in transmission loss, complex optimization models do no need to be solved in the optimization process, the influences of loads, the voltages and the reactive power are completely considered, and flexible adjusting is conducted on reactive power.