A Time Series Forecasting Model Base On ANN For Residential Price

Not only people but also government departments are most concerned about the commercialhousing prices. To research the heavy object has a certain significance.At present, the price forecast of commercial housing market is becoming a hot researchpoint , and these theory and methods are vary ,but motivation and purpose of the study are thesame, have tried to explore the invisible regular pattern of the price movements for thecommercial housing market, model, fit price trend, and complete price forecasts at last.Ningbo is a coastal sub-provincial city. Its residential commercial housing market trend hastypical, I use artificial neural network based on time series analysis method, preprocessing,modeling, learning fit, and to achieve the short-term forecasts prices of the Ningbo six districts,this short-term forecasts has some accuracy.The main work and achievements of this thesis as follows:1. Build a multi-dimensional time series of BP network topology;2. Study the application of dynamic neural network in predict real estate prices, created threemodels base on BP NN, based on RBF NN, and based on Nnstart GUI toolbox. Simulatedand predicted the Ningbo's real estate prices with actual experimental data, the resultindicated the effectiveness of the proposed method.3. Compare results of three models. Assessment of different steps impact on test.4. Summary come to the "relevance" principle, and it is helpful to the application of neuralnetwork research.