本文为土耳其比尔肯大学（作者：NURİ UÇAR）的硕士论文，共45页。

本文通过对各种线性模型和非线性模型的比较，找出对土耳其经济具有最佳通胀预测性能的模型。这些比较是通过考虑序列的类型来进行的，而不管该序列是否平稳。应用不同的组合技术提高了预测精度。结果表明，基于非平稳向量自回归（VAR）和人工神经网络（ANN）模型的组合预测优于其它模型的组合预测。此外，预测值与人工神经网络技术相结合产生的均方根误差（RMSE）比其他组合技术低。

This thesis compares a variety of linearand nonlinear models to find the one with the best inflation forecastperformance for the Turkish Economy. These comparisons are performed byconsidering the type of series whether or not stationary. Different combinationtechniques are applied to improve the forecasts. It is observed that thecombination forecasts based on nonstationary vector autoregressive (VAR) andartificial neural network (ANN) models are better than the ones generated byother models. Furthermore, the forecast values combined with ANN techniqueproduce lower root mean square errors (RMSE) than the other combinationtechniques.

1. 引言

2. 文献回顾

3. 线性时间序列的统计特性及人工神经网络模型

4. 经验结果

5. 结论

更多精彩文章请关注公众号：