34 results about "Categorical variable" patented technology

Systems and methods are provided for identifying combinatorial feature interactions, including capturing statistical dependencies between categorical variables, with the statistical dependencies being stored in a computer readable storage medium. A model is selected based on the statistical dependencies using a neighborhood estimation strategy, with the neighborhood estimation strategy including generating sets of arbitrarily high-order feature interactions using at least one rule forest and optimizing one or more likelihood functions. A damped mean-field approach is applied to the model to obtain parameters of a Markov random field (MRF); a sparse high-order semi-restricted MRF is produced by adding a hidden layer to the MRF; indirect long-range dependencies between feature groups are modeled using the sparse high-order semi-restricted MRF; and a combinatorial dependency structure between variables is output.

The invention discloses a method and a system for optimizing mobile service resources of an operator. The method includes: counting the historical dialing data of the operator's customers, and the dialing data is a continuous variable; converting the continuous variable into a discrete characteristic variable through chi-square analysis; taking whether the customer has opened a mobile service as a binary classification variable, establishing the characteristic variable and classification Variable C4.5 decision tree model, wherein, in the decision tree model, calculate the information gain rate corresponding to each segmentation, and select the segmentation threshold with the largest information gain rate as the optimal segmentation threshold for this attribute; according to the decision The tree model calculates the value of the classification variable to obtain the prediction result of whether the customer subscribes to the mobile service; optimizes the mobile service resources of the operator according to the prediction result. Through the technical solution provided by the invention, the customer's demand for mobile service can be efficiently obtained from the current customer's dialing behavior, so as to realize the optimized deployment of the operator's mobile service resources.

The invention provides a real-time prediction method of urban road traffic accident risk, which calculates by extracting the geometric linear data of each observation object in the observation set, the basic data of historical traffic flow n minutes before the occurrence of the traffic accident, and the historical weather condition data to obtain traffic The characteristic parameters of traffic flow n minutes before the accident and weather condition data are transformed into grades of categorical variables and the distribution probability of this grade, and then a real-time prediction model of urban road traffic accidents based on Poisson distribution is established, using the determined traffic flow characteristic parameters The level of weather condition data and the distribution probability of this level are used to calibrate the prediction model. When predicting the traffic accident risk of the required prediction object, it is only necessary to calculate the real-time traffic flow characteristic parameters and weather conditions of the required prediction object in real time. The level after the data is converted into a categorical variable and the distribution probability of the level can be used to predict the traffic accident risk of the desired prediction object using the calibrated formula.