An Intelligent Algorithm Based on Grid Searching and Cross Validation and its Application in Population Analysis
Yangu Zhang, Saiping Chen, Yi Wan · 2009
Population statistic and forecast is important basis that government establishes correlative policy, populationpsilas all characteristic has strong non-linear specialty because of all kinds of effects. A cross validation optimized parameter least support vector machine method of population statistic and forecast is presented aiming at bad precision and lack of rationality of all approximate model at present. Complicated and strong nonlinear population characteristic relation is simulated by network design and conformation of the least square support vector machine learning algorithm and selecting the optimized support vector machine parameters by the method of grid searching and cross validation. The model is verified by taking population growth rate for example, cross validation optimized parameter least support vector machine algorithm has strong ability of nonlinear mapping and self-learning, it avoids availably phenomenon of partial minimum and overfitting, the future population problem can be accurately calculated and judged , it gains high precision by comparing numerical value of network output with fitting value and numerical real value. It provides a new artificial intelligent approach for population analysis.