40 results about "Structural risk" patented technology

The present invention generally relates to financial data processing, and in particular it relates to credit scoring, consumer profiling, consumer behavior analysis and modeling. More specifically, it relates to risk modeling using the inputs of credit bureau data, size of wallet data, and, optionally, internal data.

The invention provides a regional people number prediction method and system based on variable components and minimum structure risks. The method comprises the steps of obtaining multiple sets of sample data of the number of people in a target region, wherein the sample data comprises a historical real user number and a historical login user variable component of the target area, and the historical login user variable component is obtained according to the historical base station login user number and the stable period user number of the target area; inputting all the sample data into a prediction model containing structural risk factors for training, and taking the prediction model with the minimum structural risk as a target prediction model; obtaining a base station login user number ofthe target area, and obtaining a login user variable component according to the base station login user number and the stable period user number; and inputting the variable component of the login user into the target prediction model to obtain a corresponding predicted value of the number of people in the target area. According to the method, the prediction efficiency and the generalization degree of the model are obviously improved while the accurate prediction of the passenger flow volume is realized.

The invention relates to a wind farm power curve modeling method based on support vector regression. First, the wind speed-power data of the wind farm is collected to obtain the upper and lower bounds of the regression curve for fitting the interval model; When the upper and lower bounds of the model need to be solved, the convex quadratic programming problem needs to be solved. According to the equivalence principle of the norm, the transformation from the second norm to the one norm is carried out, and then the linear programming problem is transformed into the one norm; then, the control model structure is The optimization problem of complexity control is applied to regression model identification, and the optimization problem of regression model identification is established by minimizing the maximum approximation error. For the optimization problem of minimizing structural risk, a polynomial kernel is introduced to obtain a new optimization problem for the upper and lower regression models. Compared with the convex quadratic programming solution of traditional support vector regression, the invention has high computing efficiency and fast solution speed.