89 results about "Lasso regression" patented technology

The invention relates to a micro-service intelligent monitoring method for abnormal propagation, which comprises the following steps of: monitoring service calling information based on an agent technology, and establishing a micro-service calling topological graph to describe an abnormal propagation relationship among micro-services; a Lasso regression modeling interface is adopted to call correlation with measurement, and abnormal micro-service is detected by monitoring the change of a correlation model; based on the PageRank algorithm, the abnormal degree of the micro-service and the callinginterface of the micro-service is evaluated, transparent service monitoring is achieved, automatic metric value prediction is achieved to find abnormal service, and the abnormal degree of nodes in agraph is intelligently evaluated to detect the problem root cause.

The invention belongs to the field of equipment fault diagnosis, and particularly relates to a sewage treatment equipment fault diagnosis system and method based on Lasso regression. The fault diagnosis system comprises a sewage treatment subsystem, a plurality of supervision clients and a plurality of data acquisition devices; the sewage treatment subsystem comprises a central console, a communication server, a total data storage server and a data acquisition server; the central console comprises a control module, a data interface module, a data classification module, a machine learning module and a data communication module; and by acquiring data on the data acquisition devices, the machine learning model carries out learning training on the data so as to construct a plurality of sub regression prediction models for predicting fault types of various types of sewage treatment equipment. The sewage treatment equipment fault diagnosis system and method disclosed by the invention providea certain basic diagnosis help for engineering supervision and treatment decisions of sewage treatment, and effectively meet requirements for real-time monitoring and diagnosis on the sewage treatment environment fault problem in the sewage treatment process.

The invention discloses a power transaction big data publishing method based on differential privacy protection, which comprises the following steps of utilizing MICFS to carry out feature selection on the correlation of the original data set of the power transaction, and selecting the data record with low correlation to generate the data set B to be determined; using a clustering algorithm to partition a K-block to obtain a plurality of sub-data blocks with mutually independent attributes; after deleting a record in the sub-data block, using the inquiry function f to inquire about the sensitivity GSD of the original data set and the sensitivity GSB of the inquiry B; according to the property of differential privacy parallel combination, adding Laplace noise, obtaining the training samplequery set satisfying the differential privacy as a set of machine learning training samples; training the Lasso regression algorithm to generate the prediction model, and inputting the original data set into the model, and outputting the query set of D. The method of the invention enables the data publishing accuracy and safety to be improved, and reduce the computing overhead and privacy budgets.

The invention relates to an improved Lasso+RBF neural network combined prediction model, and belongs to the field of big data analysis and processing. A prediction process of the model comprises the following steps of: defining life cycle features of a customer relationship so as to divide a customer life cycle into an obtaining stage, a lifting stage, a mature stage, a decline stage and a loss stage; taking customers in the loss stage as a training set and a test set of the model, taking customers in the other four stages as prediction customers, and subdividing the customers in the loss stage into the former four stages; respectively extracting features by using Lasso regression and respectively training an RBF neural network model corresponding to each stage; respectively substituting the customers in the former four non-loss stages into the trained models corresponding to the stages to carry out prediction; and finally combining the obtained prediction results to obtain a to-be-lost customer set. According to the method provided by the invention, the extracted features are more correct, the unbalance of data is decreased and the prediction accuracy is improved.

The invention relates to a market risk assessment method based on power transaction data, and the method comprises the steps: 1), carrying out the dimension reduction of a power market risk monitoringfull-index library based on a Lasso regression model, and constructing a power market risk monitoring index system; 2) determining an index threshold value by adopting a mean number principle method,screening abnormal data, and carrying out preliminary identification on a risk object; 3) calculating the subjective weight and the objective weight of each index by using an improved CRITICG1 method, calculating the comprehensive weight of each index by using an improved game theory combination weighting method, constructing an aggregation model, and determining the risk level of the screened market subject by using a comprehensive evaluation method; and 4) performing risk assessment on the market risk of the market subject according to the risk level, and establishing different early warning mechanisms to warn the market risk. Compared with the prior art, the method has the advantages of effectively monitoring potential market violation behaviors of different market subjects, improvingthe precision and speed of power market risk assessment and the like.

The invention belongs to the technical field of steel rolling product quality prediction, and particularly relates to a CNN algorithm and Lasso regression model-based hot rolling product quality prediction method. The method comprises the following steps of S1, obtaining the sample data used for modeling, wherein the sample data comprise the training data, and determining the key input variables; S2, training the CNN by using the key input variables of the sample data to obtain a feature vector model; S3, substituting the key input variables of the training data into the feature vector model to obtain an input variable for substituting into the Lasso regression model; S4, determining an optimal regularization factor of the Lasso regression model, training the Lasso regression model by utilizing the input variables in the step S3 to obtain an uncorrected mixed prediction model, and correcting the uncorrected mixed prediction model to obtain a corrected mixed prediction model; and S5, inputting the production data in the future time period into the corrected mixed prediction model to obtain a prediction result of the production data. The method can improve the prediction precision of the model.

The invention discloses a construction method of a decompensated liver cirrhosis combined infection risk prediction model. The method comprises the following steps: S1, data acquisition: collecting decompensated liver cirrhosis patient information; S2, data preprocessing: cleaning and sorting the original data; S3, index screening by adopting LASSO regression: dividing the patients into an infection group and a non-infection group according to whether the patients have combined infection or not, and performing single-factor analysis on the grouped index data of the patients to obtain single-factor meaningful indexes; bringing the single-factor meaningful indexes into Lasso regression for index re-screening to obtain indexes for constructing a prediction model; and S4, prediction model construction: constructing the prediction model by using the indexes screened by Lasso regression through multi-factor Logistic regression. The method is based on application of a clinical big data methodand is high in reliability; the constructed model is simple and easy to use, and the used indexes can be obtained through conventional inspection and are easy to obtain.

The invention relates to the technical field of water consumption prediction, and specifically relates to a water consumption prediction method and device based on big data, and the method comprises the steps: inputting collected data, and carrying out the data preprocessing through employing a principal component analysis method and a Lasso feature selection model; an annual water consumption prediction model of each region is established; obtaining a water consumption prediction result according to the preprocessed data. According to the invention, a principal component regression model is adopted; the Lasso regression model and the support vector machine regression prediction model are combined. The principal components are further screened by using a lasso algorithm after being analyzed by using the principal components. Compared with the prior art, the method has the advantages that the water consumption with high accuracy can be obtained only by collecting a small number of influence factor data sets, the prediction accuracy of the water consumption is greatly improved, and meanwhile, the water consumption is predicted within a shorter time.

Disclosed is a method for establishing a prediction model based on a multidimensional texture of brain nuclear magnetic resonance images. Images are segmented using a region growing method, a contourlet transform method is used to extract an edge texture feature parameter of ROIs, a multidimensional database is established, and a prediction model is established using various data mining methods, comprising a Gaussian process, a support vector machine, a random forest, a Lasso regression and a semi-supervised support vector machine. The ROIs comprise a hippocampus region and an entorhinal cortex region.

The invention discloses a lasso regression method for inverting particle size distribution by virtue of a light scattering method. According to the lasso regression method, the particle size distribution of particles in a measured region is obtained by virtue of known information including laser wavelength, relative index of refraction and scattered light intensity data and the like. The lasso regression method mainly aims at situations that the distribution data of the scattered light intensity in an actual measurement system can be obtained by virtue of a measuring device, and the particle sizes are unknown and need to be inverted in the dust particle detection by virtue of a light scattering method. The lasso regression method mainly aims at ill-posed problem that the search calculation in the inversion of the particle sizes for the measurement of the particle sizes is easily caught in local optimum and optical energy distribution coefficient matrixes. Under the condition of inadequate prior information, compared with a traditional inversion method, the method has the advantage that the result obtained through inversion is relatively reliable.

The invention discloses a fashion compatibility prediction method based on low-rank regularization feature enhancement representation. The method comprises the steps that a feature matrix is decomposed into a first objective function composed of main features of multiple visual angles and a sparse error matrix; The features learned in the low-rank subspace are standardized through hypergraph items, and a second objective function of the relation between the fashionable single products is obtained; a Grassmannian manifold is introduced to obtain a third objective function of the maximum distance between dictionary base matrixes under different visual angles; Establishing a relation between the characteristics of the multi-view low-rank subspace and the output matching score, adding a sparseregularization item to the least square loss part, and obtaining a typical Lasso regression, namely a fourth objective function; Obtaining a fifth objective function taking the affinity matrix as a label matrix, establishing a relationship between the affinity matrix and the learned characteristics, and minimizing an error between the affinity matrix and the learned characteristics; And obtaininga total objective function according to the weighting of the first to fifth objective functions, optimizing the total objective function by utilizing an alternating direction multiplier method, introducing a Lagrangian multiplier, and sequentially iteratively updating parameters at each view angle until the value of the objective function is converged to obtain a final prediction score.

The invention discloses a probability distribution estimation method of multi-dimensional group intelligence perception data with local privacy protection, which solves the utility and efficiency problems of multi-dimensional data in distribution estimation. Each user firstly perturbs the data at the local end and then sends the perturbed data to the central server, When the central server receives the data sent by all users, Lasso regression is used to multidimensional data with local privacy protection to eliminate redundant candidate states, And the fastest initial value of the solution based on Lasso regression is substituted into the expectation maximization algorithm, and then the convergence of the expectation maximization algorithm is applied to iterate the initial value to obtainthe accurate probability distribution estimate quickly.

The invention discloses a pruning method for an embedded network model. The pruning method is used for solving the technical problem that an existing pruning method is poor in practicability. According to the technical scheme, the method comprises the following steps: firstly, establishing a mobienet SSD network model, and carrying out a forward operation to obtain data required by pruning calculation; Channels which are not important to a convolution layer calculation result are selected through lasso regression, and channels which have relatively low influence on a summation result in the channels are selected through a lasso algorithm; The Mobienet resolves an original layer of convolution into a channel separation convolution layer and a point convolution layer, and an input channel ofthe channel separation convolution layer is equal to an output channel of the channel separation convolution layer. According to the method, the reconstruction error is reduced to serve as the core,the lasso is used for picking out unimportant channels in all convolution layers, then channel trimming is conducted on all the convolution layers according to the special structure of the mobienet, compression acceleration of the mobienet SSD is completed, and the practicability is good.

The invention provides a real-time monthly runoff forecasting method based on a deep learning model, and the method comprises the steps: 1, collecting forecasting factors based on historical information and future meteorological information, and determining the longest delay of the influence of the early monthly runoff on the forecast monthly according to the autocorrelation analysis of the monthly runoff in the historical period of a drainage basin; 2, performing normalization processing on forecast factors and monthly runoff data in a training period, and automatically screening the forecast factors by adopting an LASSO regression method based on an embedded thought; 3, clustering the training period sample set by adopting a K-means clustering method based on a division thought, and dividing samples into K classes which do not coincide with each other; 4, calculating the distance between the forecasting factor vector of the verification set and the clustering center of the K training sets, finding the nearest training set, and then training a combined deep learning forecasting model combining the convolutional neural network and the gating circulation unit network by using the data set; and 5, carrying out real-time correction on the forecast residual error by adopting an autoregressive moving average model.

The invention relates to an industrial process fault detection method based on wavelet transform and a Lasso function. The industrial process fault detection method comprises the steps of (1) obtaining normal data and fault data from a Tennessee and Eastman industrial process model, carrying out standardization processing on the obtained data, (2) carrying out wavelet transform on the normal data, compressing the normal data, carrying out Lasso regression between each set of training data processed through wavelet transform and a training data matrix in the mode that each set of training data is used as a pivot element column vector, obtaining different minimum estimated values (please see the symbol in the specification), (3) obtaining the optimal minimum estimated value (please see the symbol in the specification) through a probability density estimation method, using the optimal minimum estimated value as a threshold, and (4) sequentially carrying out wavelet transform and Lasso regression on test data, comparing the minimum estimated value (please see the symbol in the specification) obtained from each set of test data with the threshold, and judging whether each set of test data has a fault or not. Compared with the prior art, the industrial process fault detection method based on wavelet transform and the Lasso function has the advantages that all the eigenvalues are taken into consideration, and detection accuracy is improved.

The invention provides a predictor identification method of a medium-long-term runoff and a medium-long-term runoff predicting method. The predictor identification method comprises: step one, standardization processing is carried out; step two, a forecast period is set, a standardized runoff sequence Q and a climatic factor set sequence F that include a series of different lag phases form a candidate predictor set X and a corresponding standardized runoff sequence Q is used as a set Y in Lasso regression; step three, a parameter lambda is given, crossed verification is carried out and a prediction set Y' is calculated, the prediction set Y' is compared with the set Y to obtain a first evaluation index of the parameter lambda; step four, M different parameters lambda are selected, first evaluation indexes are normalized, and the results are added and then scoring is carried out; step five, statistics of total scores of the parameters lambda is carried out, and the parameter lambda withthe highest total score is selected as an optimal parameter; and step six, on the basis of the optimal parameter, regression coefficients of all climate factors obtained at the step three are obtained, wherein the climate factor corresponding to the non-zero regression coefficient is identified to be a predictor.

The invention relates to the field of non-invasive measurement, discloses method for measuring a hepatic vein pressure gradient based on portal vein characteristics, and solves the problem that a prior invasive method for measuring portal vein pressures has high risk, high cost and large operation difficulty, while the accuracy of a current clinical non-invasive prediction model is still influenced by various interference factors. The method comprises the following steps of: constructing a portal vessel three-dimensional model by using a CTA layer sequence of a liver cirrhosis portal hypertension patient, and extracting the characteristics of the portal vein; by means of LASSO regression analysis, screening out the vein characteristics closely related to HVPG, and therefore constructing the model based on portal vein characteristics to predict HVPG. The method is suitable for non-invasive measurement of the hepatic vein pressure gradient.

The invention provides a comprehensive risk scoring method for multiple myeloma, comprising the following steps: S1, obtaining a gene expression profile GSE24080 of an MM patient from a GEO database,and pre-treating genes in the gene expression profile GSE24080 to obtain the first 25% of the 5413 genes with the largest variance of expression value; S2, subjecting the 5413 genes to WGCNA gene co-expression network analysis to identify co-expressed functional modules; S3, evaluating the correlation between the functional modules and clinical information through Pearson correlation test to determine the most significant modules; S4, performing univariate survival analysis on the genes in the most significant modules by using a Cox proportional hazard model, and screening out a ten-gene scoring model composed of the 10 best genes through LASSO regression; and S5, setting the rule that each factor scores 1 point when the ten-gene scoring model or serum [beta]2M or LDH is higher than a cut-off value and otherwise 0, and establishing a comprehensive risk scoring system.

The invention relates to a ship oil consumption prediction method and device, computer equipment and a storage medium. The ship oil consumption prediction method comprises the steps: acquiring each oil consumption characteristic parameter of a ship; extracting respective oil consumption characteristic parameters by using a random forest algorithm to obtain a preset number of original characteristic parameters; preprocessing the respective original characteristic parameters to obtain target characteristic parameters; establishing an LASSO regression model according to the respective target characteristic parameters; and acquiring a current characteristic parameter, and processing the current characteristic parameter by using the LASSO regression model to obtain an oil consumption predictionvalue. According to the method, the number of the characteristic parameters required for establishing the model is reduced through the random forest algorithm, a data set with a large number of characteristics can be processed, and the data corresponding to the characteristic parameters does not need to be standardized before the random forest algorithm is used for extracting the characteristics.Through the LASSO regression model, the interpretability of the model can be improved, and the prediction accuracy can be further improved.

The invention discloses an electricity larceny prevention analysis method based on Lasso analysis. The method comprises the following steps: 1, obtaining all user electric quantity freezing data of awhole transformer area and total meter electric quantity freezing data of the transformer area from an acquisition system; step 2, carrying out data preprocessing on all data, and carrying out interpolation processing on missing data; 3, subtracting the sum of the power consumption data of all the users from the total table data of the transformer area to obtain the line loss value of each time period of the transformer area; 4, according to the Lasso regression model, calculating regression coefficients of the line loss of the transformer area and all the ammeters; 5, calculating the electricity stealing probability of each ammeter according to the line loss of the transformer area and the Lasso coefficient; 6, positioning a suspected electricity stealing user according to the electricitystealing probability. The method is easy to implement, only all user power consumption data and transformer area general table data of the whole transformer area need to be obtained, the influence ofinterference such as the needed data dimension is overcome, manual feature definition is not needed, and excessive additional equipment does not need to be added.

A defect level for a software application may be predicted by training a model using aspects of development processes from previous software applications as training data. Aspects of previous software development processes may be aggregated to form signal vectors for each deployed application. Defect scores calculated from actual defects in the deployed software applications may be paired with the corresponding development signal vectors. The signal vectors and calculated defect scores may act as training data and labels for a predictive model that uses lasso regression to generate a predicted defect score during the development process. A signal vector for a current development process may be updated in real time as the software is developed to update a predicted defect score and provide a subset of aspects in the signal vector that contribute most to the score such that actions may be taken to improve the score.

The invention discloses a two-dimensional nuclear magnetic resonance D-T<2> spectrum inversion method and device. The two-dimensional nuclear magnetic resonance D-T<2> spectrum inversion method comprises the following steps: collecting multiple sets of nuclear magnetic resonance echo string data of different echo intervals; constructing an elastic network objective function for inversion; determining a ratio c of a ridge regression regularization parameter [Alpha] and a Lasso regression regularization parameter [Beta] in the elastic network objective function; determining the optimal regularization parameter [Alpha]<2> of the ridge regression and the optimal regularization parameter [Beta]<2> of the Lasso regression in the elastic network objective function; and substituting the obtained parameters into the elastic network objective function to perform solving, so as to obtain the two-dimensional nuclear magnetic resonance D-T<2> spectrum. The invention inverts the two-dimensional nuclear magnetic resonance D-T<2> spectrum based on the elastic network, so that the inverted two-dimensional nuclear magnetic resonance D-T2 spectrum balances the smoothness and sparsity; therefore, theproblem that the inversion result is too smooth under low signal-to-noise ratio data is avoided, and a high-precision two-dimensional nuclear magnetic resonance D-T<2> spectrum can be obtained; and the invention benefits oil and gas identification and quantitative calculation.

The invention is suitable for the technical field of biology and provides a detection method and system for flora markers and a terminal. The method comprises the steps that flora sample data is acquired; according to representative sequences of strain classifications units, by adopting a system development tree algorithm, the similarity of different strain classifications units is acquired, and acorresponding similarity matrix is obtained; according to sample type markers, the abundance of the strain classifications units to which different strains belong, and the similarity matrix, througha generalized lasso regression algorithm model, a target regression coefficient vector corresponding to a set fitting effect is obtained; the strain classifications units, corresponding to nonzero coefficient elements, in the target regression coefficient vector are determined as target flora markers, and the effectiveness of the screened flora markers is improved.

The invention relates to a model construction method for predicting the prognosis risk of an early-stage colon cancer patient. The model construction method specifically comprises the following steps: firstly, obtaining a tumor tissue specimen of the patient through an operation; detecting the expression of the gene content in a tumor tissue sample of a patient by using difference analysis; then screening genes for constructing a risk model by using single-factor Cox regression analysis and LASSO regression analysis, and establishing a calculation formula of the risk model; finally, using Kaplan-Meier prognosis analysis and ROC analysis to verify the accuracy of the prognosis risk model for prognosis prediction in early-stage colon cancer patients. The method can better guide clinical design of a more targeted treatment scheme for a patient with poor early colon cancer prognosis, and realizes precise medical treatment; The number of genes for constructing the prediction model is large, the influence caused by tumor heterogeneity can be reduced to the maximum extent, and the accuracy is high; The gene detection method is simple and easy to implement.

The invention discloses a land utilization change driving force screening method and device based on LASSO regression. The method comprises the steps that explanatory variable and response variable data in a research region is acquired; a LASSO regression model is constructed, and LASSO regression is executed; a residual sum of squares and multi-colinearity indexes in each step of LASSO regressionare calculated, and driving factors needing to be eliminated are determined according to calculation results; remaining driving factors are used as explanatory variables of the LASSO regression modelto execute LASSO regression again; and the importance of the driving factors is determined according to coefficients of response variables in the LASSO regression executed again. Through the screening method and device, a refined model can be constructed, variable screening and complexity adjustment can be performed during fitting of a generalized linear model, and accurate screening can be performed according to numerous dependent variables in multi-colinearity. The screening method and device can be widely applied to the environment modeling field.

The invention discloses a research method for evaluating the cognitive performance of a driver based on driving behaviors. The method comprises the following steps: step one, carrying out data preprocessing of a collected experimental database; step two, in a regression function, selecting features by using a Pearson coefficient, constructing functions of lasso regression and rid regression, and performing cross validation to improve the accuracy and obtain predicted cognitive performance; step three, predicting cognitive behaviors of the driver through two methods; and step four, analyzing behaviors of the driver, namely analyzing an MMSE target, and performing predicting in an overall road, a residence and an intersection, especially in a residence road and an intersection.

The invention relates to a design method of an elastic network constraint self-interpretation sparse representation classifier. The method comprises the following steps: training samples are read, the training samples are linearly transformed to a high-dimensional kernel space, each type of the training samples are learnt in the high-dimensional space, a contribution (i.e., a weight) made by each individual in the type of the training sample to constructing a sub-space of the type of the training samples is found, and a dictionary is constructed by a product of the type of the training samples and a weight matrix; and elastic network coefficient coding of test samples in the kernel space is obtained through training the obtained sparse representation dictionaries, and finally, the test samples are fitted by use of each type of the dictionaries and the elastic network sparse coding corresponding to the dictionaries, fitting errors are calculated, and the type of minimum fitting errors are the type of the test samples. According to the invention, the method is integrated with the advantages of ridge regression and lasso regression, sparse coding features of the samples are enabled to sparse, the fitting errors are also quite small, classification errors are effectively reduced, and the identification performance of a classifier is improved.

The invention provides an investment benefit evaluation method based on feature extraction and lasso regression. According to the method, an investment index and power supply reliability index feature pool is constructed, an investment index associated with each reliability index is screened, a lasso regression association model is constructed based on the screened power supply reliability index and investment index combination, and the investment benefit is evaluated based on the lasso regression association model. The method can achieve precise and objective evaluation of the benefit of the investment project of the power distribution network, and guarantees high efficiency and precision of investment construction of the power grid.