75 results about "Lasso regression" patented technology

The invention discloses a hepatocellular carcinoma postoperative early recurrence prediction method based on multi-modal fusion. The method comprises the following steps: firstly, integrating clinical data of a training set, a preoperative enhanced CT image and a postoperative full-view digital pathological image, and carrying out standardized correction; then, traditional image omics features and deep learning features are extracted from the CT image, cell nucleus morphological features and tumor microenvironment spatial configuration features are extracted from the pathological image, and key feature signatures are screened out through a maximum correlation minimum redundancy algorithm (mRMR) and LASSO regression in combination with clinical features. And then carrying out progressive model construction by adopting an XGBoost algorithm, sequentially establishing a clinical single-mode model, an image single-mode model, a pathological single-mode model and a multi-mode fusion model, and explaining and visualizing the models by utilizing an SHAP value and a Grad-CAM technology. Finally, the performance of the model is evaluated in a multi-dimensional mode through internal cross validation, foresight and external independent validation, risk layering is carried out based on the prediction probability, and individualized postoperative management is guided.

The invention relates to the technical field of electricity market day-ahead prediction, in particular to an electricity market day-ahead price difference prediction method and related equipment, and the method comprises the steps: firstly obtaining electricity market boundary data, meteorological observation data and line maintenance plan data of a target area, and carrying out the preprocessing to form multi-source feature input data; and then, inputting the data into an LASSO regression model for price difference prediction. The model comprises a dynamic penalty factor generator and a near-end gradient solver. The dynamic penalty factor generator dynamically adjusts penalty factors based on multi-source features and predicted price difference in the last time period. And the near-end gradient solver adopts a near-end gradient algorithm, combines a Nesterov acceleration strategy and convergence threshold judgment conditions, iteratively solves optimal parameters, and finally outputs a day-ahead-real-time price difference prediction result. According to the method, through dynamic adjustment and efficient solution, the prediction accuracy is improved.

A time series prediction method and apparatus for the field of energy, and a medium, which relate to the technical field of energy prediction. The method comprises: collecting power generation time series data and a trend picture, and converting the trend picture into time series data; constructing and training a Lasso regression model, and saving difference time series data between a time series and a prediction result of the Lasso regression model; pre-processing the time series data and the difference time series data; inputting the pre-processed data into a time-series large model for model training; and using the trained time-series large model to perform power generation time series prediction. In the present application, preliminary prediction is performed by means of a regression model, and a time series model is used to perform difference correction on the prediction result, thereby effectively improving the accuracy of the prediction result and the interpretability of the model.

The present application provides a kind of based on machine learning's mark metabolite model prediction method of optimal harvest period of Chinese yam, steps are as follows: collecting Chinese yam samples of different harvest periods, obtain metabolomics data by analyzing Chinese yam samples through metabolomics technology;Metabolomics data are preprocessed;The feature related to the growth period of Chinese yam is obtained by using machine learning algorithm to select potential marker metabolite;LASSO regression method is used to screen potential marker metabolite to construct marker metabolite prediction model;The area under ROC curve is used to verify the constructed marker metabolite prediction model;The metabolomics data of new Chinese yam are input into the marker metabolite prediction model to obtain model score, and whether Chinese yam is suitable for harvesting is judged according to model score.The present application can accurately predict the optimal harvest period of Chinese yam, eliminate subjectivity and experience dependence, improve scientificity, reduce external environmental influence, realize Chinese yam production capacity maximization, and provide reliable technical support for agricultural production.

The invention provides a papillary thyroid carcinoma neck lymph node metastasis risk prediction method and system based on blood indexes and TI-RADS grading, and relates to the technical field of medical data analysis. According to the method, an original data set is constructed by obtaining TI-RADS classification, the maximum diameter of nodules, the number of nodules, gender, age and conventional hematology indexes including apolipoprotein B and carcino-embryonic antigen of a patient, characteristic variables are screened by adopting LASSO regression, and an independent prediction risk factor model is established in combination with multi-factor Logistic regression. And further constructing a column graph model, carrying out performance verification through multiple statistical indexes, and deploying the model to a webpage calculator based on a shiyapp to realize convenient output of the individualized risk probability. According to the method, efficient and accurate prediction of the preoperative lymph node metastasis risk is realized, and reliable decision support can be provided for clinic.

The invention belongs to the technical field of biological medicine, and discloses a portal cholangiocarcinoma postoperative risk prediction model based on a gamma-glutamyltransferase (GGT) dynamic change track and a construction method thereof. The method comprises the following steps: (1) collecting clinical characteristic data and survival characteristic data (including survival time and survival state) of a patient suffering from porta hepatis cholangiocarcinoma subjected to radical resection; 2) on the basis of GGT detection results at different time points in a perioperative period, dividing patients into different GGT dynamic change track types by using a latent category hybrid model; and 3) taking the clinical characteristic data and the GGT dynamic change track type as independent variables, taking survival data as response variables, screening prediction factors through LASSO regression, and constructing a multi-factor Cox regression model, namely a postoperative risk prediction model. According to the method, the dynamic evolution rule of the GGT is introduced, so that the limitation of single-time static detection is made up, the quantitative evaluation of the survival risk is realized, and a scientific basis is provided for individualized treatment and follow-up visit management.

The application provides a model construction and verification method and nomogram for predicting the clinical prognosis of SFTS patients complicated with AP. By collecting the clinical data of SFTS patients, the independent risk factors related to adverse prognosis are screened out by using LASSO regression analysis, a multi-factor Logistic regression model is constructed, and the model is visualized by nomogram for predicting the adverse prognosis of SFTS patients complicated with AP. The model has high discrimination and calibration, and can provide an effective prediction tool for clinicians to optimize clinical decision-making.

This invention discloses a method for predicting sudden cardiac death based on forensic autopsy data using machine learning and its forensic applications, belonging to the fields of machine learning, statistics, and forensic identification. This invention provides a method for screening independent predictors of forensic diagnosis of sudden cardiac death using LASSO regression and logistic regression. Using 10-fold cross-validation, the method selects the minimum lambda(λ) to identify 14 risk factors. Logistic regression ultimately identifies 9 independent predictors, including age, heart weight, left ventricular wall thickness, right ventricular wall thickness, interventricular septum thickness, aortic valve circumference, mitral valve circumference, liver weight, and left kidney weight. A nomogram and a web-based calculator are constructed for predicting sudden cardiac death in forensic practice, thereby determining sudden cardiac death using objective indicators. This method has not been reported worldwide.

This invention relates to the field of intrinsic ability prediction technology and discloses a method for predicting the intrinsic ability characteristics of the elderly based on machine learning. First, intrinsic ability score data of the elderly in five dimensions, including cognition and psychology, are acquired, and heterogeneity classification is performed using latent profile analysis. Using the classification results as labels, random forest and LASSO regression are used in parallel to screen key predictive factors to form a feature subset. A classification prediction model is built based on XGBoost, and hyperparameters are optimized through cross-validation and grid search. The SHAP and LIME algorithms are integrated to achieve both global and local interpretability. Finally, the model is deployed on an online interactive platform, where inputting individual characteristics outputs the IC propensity classification, probability, and feature contribution explanation. This invention achieves accurate classification and interpretable prediction of the intrinsic abilities of the elderly, reduces the dependence of assessment on professionals and the environment, and is suitable for large-scale application in grassroots elderly care scenarios.

The invention provides a multi-source feature dynamic weight wheat rust monitoring method with regional adaptability, and relates to the technical field of agricultural remote sensing monitoring. The method comprises the following steps: firstly, generating false missing points based on a buffer exclusion method, and constructing a monitoring data set with balanced positive and negative samples; secondly, combining the key phenological period of wheat, fusing multi-source data such as weather, soil, terrain and remote sensing vegetation index, and constructing an initial feature set reflecting the dynamic change of the environment; performing screening and dynamic weighting on the feature set by using a Lasso regression algorithm, and determining an optimal feature combination according to environmental differences of different regions; and finally, constructing a monitoring model by utilizing an integrated learning strategy and fusing multiple machine learning algorithms. The method can effectively solve the problems of difficulty in negative sample acquisition and poor cross-region mobility of the model in large-scale monitoring, and realizes high-precision dynamic monitoring of the suitable region of wheat rust.

The application discloses a photovoltaic array fault diagnosis method based on photovoltaic power station current signal output, collects current signals, and decomposes the denoised current signals into multiple intrinsic mode functions (IMFs) by using an improved AVMD algorithm after denoising; the improved AVMD algorithm introduces an adaptive learning rate adjustment, and combines an improved Adam algorithm to optimize and update bandwidth parameters and frequency parameters; a Lasso regression model is constructed, L1 regularization is introduced, and IMFs with higher importance are screened out as information carriers for diagnosing photovoltaic system faults; wavelet transform is performed on the selected IMFs, low-frequency components and high-frequency components on different frequency bands are obtained, coarse-grained processing is performed on the low-frequency components, multi-scale discrete entropy (MDE) calculation is performed, time scale factors and curves of normalized discrete entropy values are obtained; a threshold is set to judge the health condition of the photovoltaic system, and fault diagnosis is realized. The improved AVMD algorithm is used for decomposing and denoising the current signals, and the method has good robustness and is easy to implement.

The invention discloses a severed finger replantation postoperative necrosis risk assessment method, device and equipment and a readable storage medium, and relates to the technical field of artificial intelligence. Comprising the following steps: firstly, acquiring severed finger replantation related data of a sample patient; screening out target data of which the influence degree on the severed finger replantation postoperative necrosis risk assessment result is greater than a preset value from the severed finger replantation related data by using Lasso regression and a Boruta algorithm; then training candidate models based on the target data, and screening out a target model from the candidate models according to the prediction performance of each candidate model; and finally, obtaining target data of a target patient, and inputting the target data of the target patient into the target model to obtain an evaluation result of the severed finger replantation postoperative necrosis risk of the target patient. And the evaluation accuracy of the severed finger replantation postoperative necrosis risk is obviously improved.

The invention relates to the technical field of carbon emission, in particular to an industrial carbon emission prediction method, which comprises the following steps: acquiring and preprocessing industrial carbon emission data, removing invalid data, and calculating carbon emission and total carbon emission of energy production, basic raw material processing and manufacturing industry. Thirdly, key carbon emission influence factors are screened out through an STIRPAT model and an LASSO regression model; and then, a DDPG-RLT prediction model is constructed, future carbon emission is respectively predicted by combining a Random Forest model, an LSTM model and a Transform model, and weight configuration is optimized by using a DDPG model. And finally, modeling a Markov decision process according to future carbon emission, defining a state space, an action space and a reward function, and performing carbon emission prediction by dynamically adjusting the weight to generate a final prediction value. The method not only can improve the precision of carbon emission prediction, but also can help to realize more efficient carbon emission control and emission reduction targets in the industrial field.

The invention discloses a somatosensory tinnitus prediction model construction method and system based on clinical characteristics, and relates to the technical field of medicines.The method comprises the steps that a base line clinical data sample of a tinnitus patient is divided into a training set, a verification set and a test set, an LASSO regression algorithm is combined with a Boruta algorithm to screen out core prediction variables associated with somatosensory tinnitus from the training set, and the core prediction variables associated with somatosensory tinnitus are obtained; the method comprises the following steps: selecting a training set, performing model training and hyper-parameter tuning on the training set by using multiple machine learning algorithms, performing optimal model screening on a verification set, setting the selected optimal model as a somatosensory tinnitus prediction model, and finally performing prediction performance evaluation on the somatosensory tinnitus prediction model on a test set. The method and system provided by the invention can be used as a somatosensory tinnitus risk assessment tool.

The invention provides an agricultural net carbon sink prediction method and related equipment, and the method comprises the steps: carrying out the measurement and calculation of agricultural net carbon sink, and obtaining the amount of agricultural net carbon sink; important factors influencing the agricultural net carbon sink are screened by using a Lasso regression algorithm and an ISM algorithm, and the scientificity and reliability of a screening result are ensured; the method comprises the following steps of: constructing a data set comprising a training set and a test set by utilizing the agricultural net carbon sink quantity, important factors influencing the agricultural net carbon sink and province variables, inputting the training set into a random forest model, training a random forest, setting parameters in the random forest model according to an optimal parameter combination optimized by a grey wolf optimization algorithm, and establishing a random forest model; evaluating by using the test set to obtain a net carbon sink prediction model; and inputting the obtained important factors and province variables of the target area influencing the agricultural net carbon sink into the net carbon sink prediction model for prediction to obtain a net carbon sink prediction result of the target area, thereby improving the precision and stability of agricultural net carbon sink prediction.

The invention relates to a petrochemical engineering emergency early warning method and device based on large model data distillation. The method comprises the steps of obtaining original data, preprocessing and standardizing the original data, and constructing a data set. The method comprises the following steps: mapping a data set from a high-dimensional space to a low-dimensional space through linear transformation, and performing eigenvalue decomposition on a covariance matrix of data in the data set to obtain eigenvalues and eigenvectors; and sorting the feature vectors according to the feature values from large to small, selecting the feature vectors of which the feature value ranking is not lower than a first threshold value to form a new feature space, and extracting important features from the feature space through LASSO regression. And calling a CNN deep learning model to extract key features from the important features, distilling representative features in the original data, and outputting a feature set after distillation. And performing anomaly detection on the petrochemical engineering system through a clustering algorithm according to the feature set after distillation, and triggering early warning when a detection result does not meet a preset condition.

The invention provides a method and system for predicting the individual exposure level of air pollutants, and the method comprises the steps: S1, collecting the original sample data information of a subject population to form an original data set, and dividing the original data set into a training subset and a test subset; s2, constructing an air pollutant crowd exposure level initial model; s3, training the initial model by using the training subset, and optimizing the initial model based on the combination of a regularization method of Lasso regression, an SHAP value weighting method and a recursive feature elimination algorithm; step S4-S5, obtaining model prediction data and evaluating prediction accuracy; and step S6, collecting input characteristic variables of a target population different from the subject population, and substituting the input characteristic variables of the target population into the prediction model subjected to accuracy evaluation to realize prediction of the air pollutant individual exposure level of the target population. The prediction model is better in fitting performance and high in accuracy.

The invention relates to a nonlinear structure reduced-order model construction method based on sparse recognition and a mixed mode, belongs to the technical field of structural dynamics analysis and aeroelastic mechanics analysis, and solves the problem that complex motion caused by geometric nonlinearity under large deformation cannot be accurately described in the prior art. Comprising the following steps: S1, establishing a nonlinear finite element model of a target large flexible wing to obtain a training data set; s2, solving a mixed modal basis based on the displacement residual error and SVD (Singular Value Decomposition); s3, establishing a nonlinear stiffness coefficient solving problem model, introducing LASSO regression to establish an LASSO regression optimization objective function, and solving to obtain a sparse nonlinear stiffness coefficient; s4, based on the sparse nonlinear stiffness coefficient and the structural kinetic equation, establishing a nonlinear structure reduced-order model; and S5, applying the nonlinear structure reduced-order model to statics response solution and dynamics response solution of the large flexible wing to obtain statics response and dynamics response results.

The invention relates to the field of health behavior prediction and rehabilitation management, and discloses construction and application of a stroke patient home functional exercise compliance risk prediction model. The method comprises the following steps: collecting demographic and clinical data and scale evaluation data of a stroke patient, and carrying out missing value processing and preprocessing; identifying a compliance potential category by adopting potential profile analysis, and determining an optimal cutoff value of a total score of a scale through subject working characteristic curve analysis to form a dichotomy result; key predictive factors are screened in a training set by applying LASSO regression, and a'non-good compliance 'predictive model is established by incorporating the key predictive factors into multivariate Logistic regression; an online dynamic column graph webpage calculator is developed based on the Shiny technology, and a user is supported to input variables in real time and output a prediction probability and a confidence interval; and the discrimination degree, the calibration degree and the clinical net income are analyzed and evaluated through five-fold cross validation, Hosmer-Lemeshop test and decision curve analysis. According to the invention, rapid identification and hierarchical management of high-risk patients can be realized, and a basis is provided for follow-up visit and individualized intervention.

The invention discloses an ISMA-SVR regional carbon emission prediction method based on multi-layer influence factor screening, and the method comprises the steps: firstly accounting the carbon emission of a target region, and selecting potential influence factors related to the carbon emission of the target region for preprocessing; secondly, screening the pre-processed potential influence factors based on grey correlation analysis, and finally screening by adopting an extended STIRPAT model in combination with a Lasso regression algorithm to obtain a key feature set; and then combining a random inertia weight and a Levy flight strategy with a mucus algorithm, applying the combination to hyper-parameter optimization of a support vector regression model, and constructing a regional carbon emission prediction model. And finally, taking the key feature set obtained by final screening as input, and substituting the key feature set into the model for regional carbon emission prediction to obtain a regional carbon emission prediction result. The method is suitable for small sample scene prediction, overcomes the limitation of a single machine learning prediction model, and improves the precision of carbon emission prediction.

The invention belongs to the technical field of medical artificial intelligence, discloses an aMCI risk automatic identification system based on a polysleep electroencephalogram, and aims at solving the problems that an existing aMCI identification technology is high in subjectivity, high in cost and difficult to popularize. The system comprises a data input module, a PSG parameter automatic extraction module, an aMCI risk prediction module and a result visualization output module which are in communication connection in sequence. The data input module receives PSG original data and preprocesses the PSG original data; a PSG parameter module calls YASA, Spindler Toolbox and K complex wave detection methods, and nine core parameters are automatically extracted; the risk prediction module outputs aMCI risk probability through LASSO regression and column diagram models; and the result module generates a PDF report, is in butt joint with a medical platform and stores historical data. The system realizes full-process automation, predicts AUC to be more than 0.95, adapts to multiple scenes, and provides an accurate and economical tool for early recognition of aMCI.

The invention provides a construction method and system of an acute obstructive suppurative cholangitis conservative treatment failure risk prediction model. The construction method comprises the following steps: S1, collecting clinical and laboratory index data of a plurality of cases of acute obstructive suppurative cholangitis patients conforming to a containing and discharging standard, preprocessing the clinical and laboratory index data, and then dividing the clinical and laboratory index data into a training set and a test set; s2, for the data of the training set, balancing is carried out by adopting an ROSE algorithm, and then feature screening is carried out by sequentially utilizing Spearman correlation analysis, single-factor ROC analysis and LASSO regression; s3, taking whether the patient is aggravated or not within 24 hours as a dependent variable, taking the features as independent variables, establishing a prediction model by adopting multivariable logistic regression, and optimizing model parameters through 10-fold cross validation; and S4, performing model performance verification evaluation based on the test set. The early prediction precision is significantly improved, and risk layering and treatment decision quantification are realized.

The invention relates to the technical field of plough layer structure identification and evaluation, in particular to a reasonable plough layer evaluation index system construction method based on dual-objective optimization. Comprising the following steps: acquiring target area cultivated land multi-source soil data, and preprocessing to obtain an original soil attribute variable set; constructing a double-target comprehensive scoring function to obtain a comprehensive scoring value to divide the levels of the plough layer; processing the original soil attribute variable set by adopting a principal component analysis method to obtain a first variable set; constructing an LASSO regression model to screen the first variable set and the original soil attribute variable set to obtain a second variable set and a third variable set; acquiring a fourth variable set by adopting a random forest algorithm; key variable sets of three levels are determined through intersection comparison, and a reasonable plough layer level and index system is constructed in combination with the plough layer level of the cultivated land in the target area. According to the method, collaborative quantification of a plough layer output function and an anti-corrosion function is achieved, and a scientific basis is provided for farmland quality improvement, soil improvement and partition management.

The application provides a method for constructing a PCI postoperative patient all-cause death risk prediction model, which comprises the following steps: obtaining a training data set containing clinical electronic medical records, blood cell count and blood biochemical data of patients undergoing selective PCI, and at least 5 years of postoperative follow-up and all-cause death label; performing Z-score standardization on continuous variables and one-hot encoding on category variables for pretreatment; screening 43 core features through LASSO regression combined with ten-fold cross-validation; constructing a model based on a LightGBM algorithm, taking a binary logarithmic loss function as a measurement standard and optimizing hyperparameters through grid search; finally outputting a prediction model which can receive core feature data of a new patient and output the postoperative all-cause death probability of the new patient; and the application can realize accurate prediction of long-term death risk of PCI postoperative patients.

The invention relates to the technical field of biomedicine, in particular to a group of key biomarkers for predicting endometrial receptivity, including specific microbial strains and metabolites, and a prediction model containing the markers. The method comprises the following steps: collecting a uterine cavity lavage fluid sample, carrying out 16S rRNA gene sequencing and non-targeted metabonomics analysis, and screening out eight key microbial strains and eight key metabolites; based on the markers, the following diagnostic models are constructed: a model only based on eight microbial strains; a model based on only eight metabolites; a combined microorganism and metabolite model; and a clinical data model is further combined. According to the method, a microbiome and metabolome multi-omics conjoint analysis strategy is adopted, strict screening is carried out in combination with LASSO regression, SVM-RFE and other machine learning algorithms, and the obtained biomarker is high in specificity. The constructed model, especially a combined model, shows high prediction precision and good calibration degree through verification of various curves.

The invention provides a method, a device, equipment and a medium for predicting the expression level of renal cancer HIF-2alpha. The method comprises the following steps: S10, acquiring clinical pathological data and an abdominal enhanced arterial phase image of a patient with renal clear cell carcinoma; s20, sketching a three-dimensional region of interest of the tumor through the abdominal enhanced arterial phase image; s30, extracting image radiomics characteristics from the tumor three-dimensional region of interest, screening the image radiomics characteristics through Z-score standardization processing, stability test, an mRMR algorithm and an LASSO regression algorithm in sequence to obtain screened characteristics, and calculating to obtain a radiomics score through the screened characteristics; s40, screening out independent clinical predictive factors related to HIF-2alpha expression from the clinical pathology data, wherein the independent clinical predictive factors comprise WHO / ISUP classification and clinical T classification; and inputting the independent clinical prediction factor and the radiomics score into a joint prediction model to obtain a prediction score, and evaluating the probability of HIF-2alpha expression by using the prediction score.

The invention discloses a method and equipment for determining a pipeline cleaning strategy, a medium and a product, relates to the technical field of pipeline oil stain cleaning, and aims to solve the problem of how to optimize the pipeline cleaning strategy. The method for determining the pipeline cleaning strategy comprises the following steps: acquiring sample data of pipeline cleaning; based on the sample data, obtaining a cleaning effect multiple regression model, a cleaning effect ridge regression model, a cleaning effect lasso regression model and a cleaning effect stepwise regression model; determining an error evaluation index of the pipeline cleaning strategy determined by each regression model; determining a target cleaning rate model from the cleaning effect multiple regression model, the cleaning effect ridge regression model, the cleaning effect lasso regression model and the cleaning effect stepwise regression model, wherein the target cleaning rate model determines that an error evaluation index of the pipeline cleaning strategy is smaller than a preset index threshold value; and determining a target strategy for pipeline cleaning through the target cleaning rate model.

The invention discloses a pavement performance prediction method for a rural simple pavement, and the method comprises the steps: data standardization: constructing an original matrix based on a disease evaluation value, and carrying out the standardization; carrying out formalized quantitative analysis, collecting disease, climate and traffic data, calculating weights and road surface damage condition indexes through a dynamic weight function, and carrying out multi-level summarization; constructing a rural highway use performance evaluation system, and performing PQI evaluation based on sampling data; feature selection, systematic analysis of disease types, and screening of main diseases through optimal subset regression, LASSO regression and random forest; index entropy and weight are calculated, disease entropy is quantified and standardized, and entropy weight is calculated based on information utility value. According to the method, the problems that in the prior art, rural simple paving materials lack systematic research, maintenance intervention influences are ignored, disease weight distribution is not scientific, and classification depends on artificial experience, so that model generalization is poor are solved, and the scientificity and accuracy of rural road maintenance are improved.

The application discloses a kind of career assessment method and system based on recruitment big data, comprising: collecting relevant recruitment information;The recruitment information is carried out data preprocessing, obtains the data set corresponding to the recruitment information;Text mining is carried out to the data set, obtains the keyword that has influence on average salary and the average salary corresponding to the keyword;The information of the job applicant is substituted into the Lasso regression model for salary prediction constructed in advance, and the salary expected value of the job applicant is obtained;According to the keyword and the regression coefficient of the Lasso regression model, the career assessment scheme of the job applicant is determined;The career assessment scheme is used to carry out career assessment to the job applicant. Solve the problem that the job seeker cannot reasonably position oneself and cannot understand market demand in time.

This invention discloses a method for screening key lipids during buckwheat storage based on lipidomics and machine learning, belonging to the field of food quality detection and metabolic biomarker screening technology. The method includes: collecting buckwheat samples under different storage conditions; obtaining volatile compound data through GC-MS analysis; acquiring lipidomics data of the buckwheat samples using lipidomics analysis; measuring quality-related indicators of the buckwheat samples; preprocessing the volatile compound data and lipidomics data; constructing a LASSO regression model to extract lipid molecules with non-zero coefficients as the first candidate set of key lipids; constructing a random forest regression model to calculate the importance score of each lipid molecule; and evaluating the predictive performance of key lipid molecules on the changing trends of marker volatile compounds. This invention is the first to combine LASSO regression and random forest regression for screening key lipids during buckwheat storage, overcoming the limitations of single methods, resulting in more scientific and reliable screening results, and providing a new method for revealing the molecular mechanism of buckwheat storage quality formation.