用于研究时间序列的方法有AR（自回归）、MA（滑动平均）、ARMA（自回归滑动平均）这三种模型。而对于一个平稳时间序列预测问题,首先要考虑的是寻求与它拟合最好的预测模型。而模型的识别与阶数的确定则是选择模型的关键。

1.用 迭代生成1000个点（前2个点自定义）。

2.在这1000个点中取800点进行时间序列分析建立合适的模型。

3.利用剩余的200个点进行模型预测，并看其是否匹配，最后校正。

-Methods for studying time series are AR (autoregressive), MA (moving average), ARMA (autoregressive moving average). For a stationary time series prediction problem, the first thing to consider is to find the best prediction model with it. And the identification of the model and the order of the selection is the key to the selection model.

1. Use iteration to generate 1000 points (the first two points to customize).

2. Take the 800 points in the 1000 points for time series analysis to establish the appropriate model.

3. Use the remaining 200 points to model the forecast and see if it matches and finalize it.