131 results about "Probabilistic forecasting" patented technology

Advertisement response probabilities are utilized to alter advertisement scores. A plurality of possible advertisements is accessed from, for example, an advertisement database or advertisement pipeline. A response probability for each advertisement is determined. A response probability may be a probability that a user will “click,” or otherwise select an advertisement. Advertisements may be associated with probabilistic prediction models that take advertisement recipient attribute values as inputs and provide a probability distribution as output. A score associated with each of the possible advertisements is altered based on the response probability for each of the advertisements. Statistical prediction is used to determine how scores are to be altered. Advertisements with response probabilities less than a mean probability may have associated scores decreased. Conversely, advertisements with response probabilities greater than a mean probability may have associated scores increased.

Prediction markets are used to determine the probability of an experimental therapeutic, diagnostic, or prophylactic candidate meeting clinical trial and post-trial goals, such as clinical trial endpoints and timelines. The prediction market processes buy and sell orders from market participants, while adjusting the prices of the securities according to the orders. The securities have specific meanings which correspond to goals in clinical trials or other outcomes in clinical candidate development. The price of a security determined by the market corresponds to the probability of the corresponding goal or outcome. The participants are selected for their expert knowledge of specific factors related to candidate development. Using appropriately selected securities and participants, the prediction market may be used to generate probabilities of success useful for long-range planning and valuation, determining production timelines and volumes, management of candidates in a development portfolio, and clinical management of patients by physicians.

A system and method for making a decision of whether to carry additional fuel on an aircraft for a particular flight based on a forecast, such as for low visibility and ceiling. Preferably, observations-based probabilistic forecasts are utilized. The forecast probability of the weather at the planned aerodrome being below a prescribed minimum level is calculated using statistical regression analysis of past data. An optimal probability is estimated using cost parameters on an individual flight bases. If this forecast probability is greater than the optimal probability for a particular flight, then extra fuel is carried by that flight. This is in contrast to current practice whereby the same categorical forecast is applied to all flights. The combination of improved short-term forecasts and identification of optimal forecast probabilities minimizes the financial impact of errors and weather forecasts on airline operations thereby providing a superior financial outcome.

The invention discloses a language identification method of a scene text image in combination with global and local information. Basic features of a character image are extracted, and then global andlocal feature representations are extracted respectively; the global extraction branch uses global maximum pooling to express the whole graph as a vector, and category score prediction is carried out;probability prediction is performed on the local blocks of the image by the local aggregation branches respectively, and then the series of probability distributions are combined to obtain a categoryprediction score of a local level; and finally, global and local prediction scores are dynamically fused according to the branch prediction conditions to obtain a final identification result. According to the method, overall features and local differentiated features of the character images are noticed at the same time, and end-to-end training can be achieved in one step. Compared with an existing technology utilizing local features, the method has the advantages that the local differentiated features can be accurately extracted, excellent effects are achieved in the aspects of accuracy, operation efficiency and universality, and high practical application value is achieved.

An embodiment of the invention provides a repayment probability prediction model building method and device, and belongs to the field of credit prediction. The method includes the steps: firstly, acquiring a plurality of default customer historical data sets as training sets of a model; secondly, acquiring a loan customer data characteristic set from the training sets by characteristic extraction;building a repayment probability prediction model based on the loan customer data characteristic set and a preset algorithm. The repayment probability of customers can be predicted by the built repayment probability prediction model, so that the customers with lower repayment probability can be effectively recognized, high-risk customers are discovered as early as possible, repayment is collectedby corresponding collection means, more reasonable collection arrangement is realized, and collection efficiency is remarkably improved.

Systems, methods, and computer program products that enable system-wide probabilistic forecasting, alerting, optimizing and activating resources in the delivery of care to address both immediate (near real-time) conditions as well as probabilistic forecasted operational states of the system over an interval that is selectable from the current time to minutes, hours and coming days or weeks ahead are provided. There are multiple probabilistic future states that are implemented in these different time intervals and these may be implemented concurrently for an instant in time control, near term, and long term. Those forecasts along with their optimized control of hospital capacity may be independently calculated and optimized, such as for a dynamic workflow direction over the next hour and also a patient's stay over a period of days. In the present application, a probabilistic and conditional workflow reasoning system enabling complex team-based decisions that improve capacity, satisfaction, and safety is provided. A means to consume user(s) judgment, implement control on specific resource assignments and tasks in a clinical workflow is enabled, as is the dynamical and optimal control of the other care delivery assets being managed by the system so as to more probably achieve operating criteria such as throughput, waiting and schedule risk.

The invention relates to a power system load prediction method based on a Markov chain. Under the condition that a value Lt-1 is known, according to historical data, various change trends of next time t are counted, probabilities are counted, and a trend with a largest probability is taken as a final prediction result. The method has the following advantages: load prediction can be carried out with a few samples, operation speed is fast, operation time is short, and a result of probability prediction can be obtained.

The invention relates to a deep neural network in artificial intelligence, and provides a default probability prediction method and device, computer equipment and a storage medium. The method comprises the steps of acquiring an insurance source identifier, and inquiring whether a matched target entity exists in a preset knowledge graph or not according to the insurance source identifier; When thematched target entity exists, obtaining a to-be-predicted entity associated with the target entity in a preset knowledge graph; And based on the state attribute information of the to-be-predicted entity and the association relationship between the to-be-predicted entity and the target entity, obtaining a risk early warning feature vector, inputting the risk early warning feature vector into a trained risk early warning model for prediction, obtaining a default probability vector, and obtaining the default probability of the to-be-predicted entity according to the default probability vector. Byadopting the method, the efficiency and accuracy of default probability prediction can be improved.

The invention discloses a method and system for issuing overseas coupons based on estimated income. The method for issuing overseas coupons includes: acquiring user historical data; processing the user historical data through XGBoost algorithm respectively, and establishing a user order amount prediction Model; predict the amount of the user's order; process the historical data of the user through the XGBoost algorithm respectively, and establish a prediction model for the probability of the user's order; predict the probability of the user's order; Estimated revenue of coupons, issuing coupons corresponding to the maximum estimated revenue to users. The present invention overcomes the defects of issuing coupons in the prior art, and realizes the purpose of maximizing user benefits while stimulating users to place orders.

The invention relates to an unstructured data default probability prediction method based on deep learning. The method comprises the steps as follows: unstructured data, including text data and time series data, of credit subjects are integrated and cleaned; the unstructured data are converted into a data format recognizable by a deep learning model; data features are extracted as sample data on the basis of a deep learning model frame; as for the extracted sample data, a credit risk model is constructed by use of a complex machine learning classification algorithm-integrated tree model, and default probability prediction is output. According to the method, the unstructured data such as text and time sequence data are mined, potential risk behavior modes of the credit subjects are caught on the basis of deep learning and a big data technology, high dimensional data credit risk modeling is performed accordingly, automatic, comprehensive and procedural quantitative credit risk analysis for the credit subjects is realized, the finance risk control capacity is improved, and the credit risk is reduced.

A system and method for the automated generation of documents for a legal transaction over a network using probabilistic prediction of customary usage. The predictions are generated by user experience, expert rules and machine learned classifiers based on user input of transaction data. The classifiers are constructed and tested on a partitioned dataset consisting of transaction data and legal document and clause selections in previous transactions. In one embodiment, such dataset is collected in a document management system.

An embodiment of the invention discloses a method and a device for phonetic annotation of Chinese characters. The method comprises steps as follows: when polyphone fields in an input text are subjected to phonetic notation, at least two pronunciations of the polyphone fields are acquired; probabilistic forecasting is performed on each pronunciation of the at least two pronunciations, and a forecasting result is generated; and a current pronunciation of the polyphone fields is determined according to the forecasting result. Accurate phonetic notation of polyphone words can be realized, and the text reading efficiency and effect are improved.

The invention discloses a text classification method combining a title and a text attention mechanism. The method comprises the following steps: firstly, carrying out word segmentation preprocessing on a title and a main body of each document to obtain a title word set and a main body word set; training a word vector by adopting a word2vec CBOW model, the expression of each word combined with context semantics is learned by using a bidirectional recurrent neural network, and the potential semantic vector of one word is obtained through serial word vectors and the expression of left and right contexts of the word vectors; respectively carrying out maximum pooling processing on the potential semantic vectors of each word in the title word set and the text word set to obtain a title vector and a text vector; obtaining an attention vector by using a title and text attention mechanism; and after the vector representation of the whole document is calculated, outputting the category of the probability prediction text through a softmax function. The method solves the problem that the classification result is low in accuracy because the importance of the title content is ignored and the title is taken as one part of the text or the title information is ignored in the existing text classification with the title.

The invention discloses an establishment method of a solitary pulmonary nodule malignancy probability prediction model. The establishment method particularly includes the steps: acquiring basic information of patients and serum tumor marker levels 1-7 days before operation; dividing patient cases into one group with GGO (ground glass opacity) lesion proportion higher than or equal to 50% and another group with GGO lesion proportion lower than 50% according to the GGO lesion proportion and CT (computed tomography) imaging reports of the patients; setting experiment groups and validation groups in each group of cases according to the proportion of 3:1, performing single-factor analysis on relative data of cases of the experiment groups to initially screen independent risk factors; substituting the independent risk factors into multifactor analysis to obtain independent risk factors for judging benign and malignant SPNs (solitary pulmonary nodules); acquiring the SPN malignancy probability prediction model by the aid of Logistic regression; substituting case data of the validation groups into the model, and verifying the case data of the validation groups. The model is simple and easy to use, used indexes can be acquired by the aid of routine examination and are easy to use, and effective intermediate reference information can be provided for further diagnosis and treatment of doctors according to the model.

The invention discloses a wind power climbing event probability prediction method and system based on a Bayesian network, and the method comprises the steps: mining the dependency relationship betweena wind power climbing event and related meteorological influence factors such as wind speed, wind direction, temperature, air pressure, humidity, and the like, and building a Bayesian network topological structure with the highest fitting degree with sample data; quantitatively describing a conditional dependency relationship between the climbing event and each meteorological factor, estimating the value of each conditional probability in a conditional probability table at each node of the Bayesian network, and forming a Bayesian network model for predicting the wind power climbing event together with a Bayesian network topological structure; deducing the probability of occurrence of each state of the climbing event according to the numerical weather forecast information of the mastered prediction time; the value of the corresponding conditional probability at each node is adaptively adjusted, so that the inferred conditional probability result of each state of the climbing event is optimized, and the compromise between the reliability and the sensitivity of the prediction result is realized.

The invention provides a real estate customer transaction probability prediction method and device and a server, and the method comprises the steps: obtaining the historical behavior data of a to-be-tested customer for a target building, the historical behavior data of the to-be-tested customer comprising one or more behavior characteristics and a corresponding first occurrence frequency; obtaining a decision tree structure obtained by training the training set by using a random forest algorithm model; wherein the training set comprises historical behavior data of a client selected from the database and a client type, and the historical behavior data of the selected client comprises one or more behavior characteristics and a corresponding second occurrence frequency; inputting the historical behavior data of the to-be-tested client into the decision tree structure to obtain the transaction probability of the to-be-tested client to the target building; therefore, the effective prediction of the customer transaction probability is realized, the intention analysis difficulty of the sales personnel on the real estate customer transaction probability is reduced, the sales personnel canconveniently carry out more targeted follow-up service on customers with different transaction probabilities, and the sales performance is improved.

The invention discloses a wind power probabilistic forecasting method based on a longitudinal moment Markov chain model. Wind power historical data of corresponding moments are recorded through reasonable longitudinal moment and state partition to form a longitudinal moment Markov chain probability transfer matrix which embodies state transition probability characteristics of longitudinal moments; on the basis of wind power probabilistic forecasting results, forecast is carried out by recording the probability distribution of wind power output variation of an adjacent moment and setting expectation of a confidence interval over probabilistic forecasting, probabilities exceeding the confidence interval are amended through the expectation of variation, and precision of determinacy predicted values is improved. Through verification of actual wind field data and comparison of various error indicators, the effectiveness of the model and the forecasting method is confirmed.

The invention relates to a Markov chain based method for accurately forecasting power system loads. The method is characterized by counting various variation trends of the next moment t according to the historical data and counting the probability of the variation trends under the condition of the given value Lt-1, finally taking the trend with the highest probability as the forecast result and obtaining the final result by correcting the sum of the probability on the two sides of the forecast result. The method has the following advantages: the loads can be forecast by only a few samples; the operating speed is high; the operation time is short; and the accurate probability forecast result can be obtained.

The invention discloses a routing method of social perception and probability prediction for a mobile opportunity social network. According to the method, two kinds of related centrality are provided in an initialization phase by researching the liveness, interest and mobility of nodes and the relationship between the nodes so as to simplify a network model; and in a routing phase, message transmission is carried out on the basis of comprehensive consideration of social perception and probability prediction depending on the simplified network model. In addition, a message transcript control strategy is adopted in the message transmission process to effectively control the number of transcripts of a message in the network and reduce the consumption of network resources. The simulation experiment results show that compared with the other three classical routing methods, the method disclosed in the invention has the advantages of effectively improving the delivery rate of the message and reducing the overhead rate of the network.

The invention discloses a wind electricity probability prediction-based dispatch demonstration method and belongs to the electric power industry regulation technical field. The method includes the following steps: S1, establishing a wind electricity power probability prediction model based on component sparse Bayesian learning, and obtaining prediction information according to the wind electricity power probability prediction model; S2, establishing a model constraining electric power probability dispatch upper limit and considering system risks according to the prediction information, and formulating a power generation plan arrangement strategy according to the result of an electric power system probability dispatch model; S3, performing wind electricity probability prediction according to the power generation plan arrangement strategy, and obtaining and demonstrating the result of wind electricity probability prediction; and S4, realizing wind electricity probability prediction-based dispatch according to the demonstration result. With the method the invention adopted, the errors of short-term wind field output power value prediction values can be decreased, and the accuracy of power generation plans can be improved, and risks that power generation of a power system does not satisfy loads can be eliminated, nearly 10% of wind electricity absorption volume can be increased, and the economic benefits of power grid enterprises can be increased.

The invention discloses a backoff method based on probability forecasting. According to the method, by establishing a Markov chain model, the relations of the changes of collison probability and expiring probability of a beacon message along with the minimum contention window are deduced theoretically, and it is guaranteed that the acceptance rate of a broadcast message achieves the optimum; with regard to the problems that backoff factors of a traditional vehicle-mounted self-organizing network backoff method are too simplex, and backoff order number magnitude is overlarge, the backoff method based on probability forecasting (CEB for short) is provided. The method is put forward on the basis of the idea of the relations among beacon message collision probability, the expiring probability and the competition window values, and the competition window values are adjusted according to the relative magnitude and linearity according to the collision probability and the expiring probability. Through the method, the acceptance rate of the broadcast message and performance in the average arrival time delay are improved, the fairness of channel access is optimized, and the priority of an access channel of the emergent broadcast message can be guaranteed.

The invention discloses a method for combining logistic regression credit approval based on user data and expert features. The method comprises the steps of inputting data cleaning, carrying out datadimension reduction and preprocessing, carrying out data classification, data feature engineering and feature extraction, introducing expert features, carrying out feature prediction and outputting anapproval list. In the present invention, according to the credit approval method, expert features in a traditional financial model are combined with a classic machine learning method. The future default possibility of possible dynamic change is predicted by combining market real-time updating data and feature engineering. A prediction model and an optimized logistic regression algorithm are adopted. The complex credit constraints are met. Obtained default probability prediction and risk premium results are more accurate. Auditors can be liberated from heavy credit risk assessment auditing andpricing. The large-scale small and micro enterprise credit approval can be rapidly achieved. It is possible to guarantee intelligent rating and avoid risks.

The present invention provides a probabilistic forecasting method of a coastal gale caused by a tropical cyclone. The method comprises a step of collecting historical tropical cyclone data which has a gale influence on an area to be measured and hour instantaneous maximum wind speed history data of the area to be measured at a corresponding time, a step of classifying the historical tropical cyclone data according to intensity to obtain the data of various types of tropical cyclones, a step of dividing the forecasting range of the area to be measured into grids, and calculating the probability of the gale of the area to be measured caused by tropical cyclones in the grids and drawing the gale probability distribution map of the area to be measured caused by the various tropical cyclones, and a step of predicating and outputting the gale probability of the area to be measured combined with the gale probability distribution map according to the strength information and path information of a tropical cyclone to be measured in the future. The invention also provides a corresponding probabilistic forecasting system.

The invention discloses a wind power combination probability prediction method considering evaluation index conflicts. The method is characterized by comprising the steps of determining a decomposition parameter K through variational mode decomposition optimized on the basis of the law of conservation of energy, decomposing an original wind power signal into a series of intrinsic mode function components, removing an intrinsic mode function with the minimum amplitude, and combining the remaining intrinsic mode functions to obtain a wind power sequence after fluctuation and randomness are reduced; constructing an input feature set containing 96-dimensional historical features by using the wind power sequence, and constructing different GPR models by using 10 covariance functions; calculating the area grey correlation closeness based on the five indexes by adopting an area grey correlation decision-making method so as to comprehensively evaluate the performance of each prediction model and solve the conflict between evaluation indexes; and calculating the weights of different GPR probability prediction models in the combined model according to the area grey correlation closeness, constructing the combined model, and carrying out wind power probability combined prediction by using the combined probability prediction model.