This is the repository of time series forecasting models modified by xinze.zh.
Based on the work of other researchers, this repository is expected to make the codes easy-reading as well as providing more succinct, standard, easy-to-use API that can forecast time series.
This is a pytorch-based modification of DeepAR (DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks).
Based on the works of TimeSeries, this implementation is developed to provide more easy-reading API for training and testing single variable as well as multi-variable time sries.
- Yunkai Zhang, Qiao Jianga, and Xueying Ma who are the original authors of TimeSeries.
This is a pytorch-based modification of ConvRNN(Autoregressive Convolutional Recurrent Neural Network for Univariate and Multivariate Time)
- KurochkinAlexey, Fess13 who are the original authors of ConvRNN.
This is a numpy-based Multiple-output support vector regression which implements support vector regression with multi-input and multi-output. This package is based on our group's paper Multi-step-ahead time series prediction using multiple-output support vector regression. The details of usage can be found in MSVR.
- Xinze Zhang, Kaishuai Xu, Siyue Yang who are the original authors of MSVR
- This work was done under the direction of our supervisor Prof. Yukun Bao.