Comprehensive Study Of Statistical Forecasting Methods And Computer Realization

Forecasting is to predict the development of things in the future, its application is very extensive. As one of the most effective means to deal with data, statistics is playing a important role in forecasting. We consider statistical forecast as one kind of method or science for getting the forecasting results with statistical methods and so on, which is based on quantitative analysis . Nowadays, the study of statistical forecasting methods is going deeper and deeper. And its relationship with other sciences is going closer and closer. This makes people more and more difficult to master the statistical forecasting methods. However its application is quite extensive.To solve this problem, the best approach is to realize these methods via computer and offer an effective tool. After making a general survey of the relevant software both at home and abroad, we find some shortage. This will become the starting point of our paper. We hope that we could develop an ideal statistical forecasting software. During the development of the software, the most difficulty is how to deal with the data and get the effective algorithm. Against such a difficulty, we raise a creative method which processes the data as matrixes. We have worked out the corresponding class of matrix and its operation storehouse via computer. This method will solve this problem in a good way.During the course of further study, we found some questions. We thought it over corresponding with these questions and raised our views. These include the folio wings: l.we raised the concept of direction-error; 2.we considered the forecasting errors and calculated the right of the comprehensive forecasting models with AHP; 3. we analyzed some examples and gave some helpful suggestions about the selection of parameter in the exponential smoothing methods; 4. we thank it over with the using of data during the course of predicting and so on.At last, we introduce the developing directions of further research of statistical forecasting methods with some documents. We introduce two points of them in detail and this will involve some interesting questions.