Prediction Algorithm Research And Applications

In this paper, we mainly discuss several forecasting methods, as well as their applications in Zhejiang electricity markets and Chinese stock markets. Here we briefly introduce statistic methods and BP algorithm. We make analyses of the three hypotheses of regression . adaptive adjustment of smoothing index parameters and parameter estimation of ARMA models. After that, we discuss relations and differences between BP and statistic algorithm. Then, some methodologies on data clearing are introduced, and we propose an algorithm on variable selection through sensitivity analysis.Reforms of power industry have made market clearing price (MCP) forecasting more important. This article presents a MCP forecasting method according to Zhejiang circumstance. Application to the real power system shows the accuracy of the proposed method is about 90%. On the base of MCP predictions, we study generation companies' bidding strategies, and implement corresponding software. In this section, a clustering method based on Adjacency Matrix of Graphs is proposed.Forecasting is not always effective for any applications. The fourth section of this paper presents such a case. We use linear and nonlinear technologies to analyze profit of Chinese stock markets, and the results shows that one can' t expect to get any valuable information of profit in the future by mining historical data. This indicates Chinese stock markets possess weak validity.