The Deduction and Application of Climate Index Based on the L-M BP Neural Network Algorithm in Chinese Real Estate Market

The paper introduces the intelligent neural network to make forecast for the climate index in the real estate market. At the same time, deeply mine the practical use of intelligent system of neural network, thus, abundant chaos phenomenons are obtained in the hidden nodes by numerical simulation. The bifurcation diagram and chaotic attractors are drawn from two dimensional chaotic neurons to seven dimensional neurons, and the sensitivity of initial weights for neuronal cell is verified. By comparing the traditional BP neural network with the improved L-M algorithm, we obtain the conclusion that the parameters of the nonlinear climate index system can affect the output of climate index system because certain ranges of parameters of hidden nodes may cause chaos. However, the improved L-M algorithm can choose the best parameters for the system, so that the complexity of the climate index system is reduced and chaos disappears, in addition it can shorten the training time, getting the higher precision and the faster convergence speed compared with conventional BP neural network in the system of climate index, which is better for the evaluation of climate index in the real estate market. Key–Words: The traditional BP neural network, Improved L-M algorithm, Climate index, Highdimensional hidden nodes, Bifurcation diagram, Chaotic attractors.