Enterprise Management Forecast Method Study Based On Wavelet

With the high-speed development of market economic and constantly and intensely competition, the leaders of enterprise more and more recognize that it gets nowhere if they only depend on traditional and experiential decision-making, so they extremely take count of how to improve decision-making methods. The correct and reliable forecast and the decision-making come from it are the key factors for company to get success . The wavelet method used in economic forecast depends on its "mathematics microscope" property. It does the layer analysis and forecast to indicators. It can improve forecast precision, what's more it can search and express the structural feature of data such as development cycle, second cycle , especially to some sudden change data which will provide effective and reliable warranty to the complexity and violent fluctuant data indicators in enterprise.The thesis is one part of Beijing scientific research subject which name is "economy forecast and decision-making methods study" , object of which is to provide a set of new time-frequency forecast methods based on wavelet theory and multiple dimensions thinking way which is not the same as the formerly time methods or frequency methods to help the enterprise administrator to do the multiple angle and high quantity forecast work.The thesis researches and explores the enterprise management forecast methods base on the Matlab tool, Eviews software and compute. The process of noise elimination based on wavelet is to eliminate disturbed noise of enterprise signals which produced when it bored, conveyed and processed. The elimination noise process is to use the valve filter method to filter the high frequency coefficient, and reconstruct the low frequency and high frequency which noise have been eliminated, the signal after noise eliminated is more glossy and it is helpful to do the trend forecast It can analyze and do trend forecast to the total trend, periodic and sudden change of signals when do the trend forecast to the enterprise signals. The first process is to select the wavelet functions and maximum decomposition dimensions, then use the combination of wavelet decomposition and wavelet reconstruct to do the forecast to the signals which is be researched. The forecast model based on wavelet and B-J method is to use the model to forecast after simplifying the enterprise signals without energy losing. It can explicit the mechanism of the process of signal basically, and determinate the action power and develop trend of the factor which is as a result of the signals' structure, then attain to change or maintain the development of signals. The enterprise development indicator is given from the requirement of enterprise develop decision-making. It is a synthetic indicator combined of single indicator. It can depict the development level of enterprise quantify. Synthetic forecast indicator is more complicate than single indicator. The effective combination of wavelet and NN can increase the forecast precision and provide the guidance to thedecision-making of enterprise.