Deep State Space Models for Time Series Forecasting
Authors: Syama Sundar Rangapuram, Matthias W. Seeger, Jan Gasthaus, Lorenzo Stella, Yuyang Wang, Tim Januschowski
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We provide qualitative as well as quantitative results with the proposed method, showing that it compares favorably to the state-of-the-art. |
| Researcher Affiliation | Industry | Syama Sundar Rangapuram Matthias Seeger Jan Gasthaus Lorenzo Stella Yuyang Wang Tim Januschowski Amazon Research {rangapur, matthis, gasthaus, stellalo, yuyawang, tjnsch}@amazon.com |
| Pseudocode | No | The paper describes the computational procedures involved (e.g., Kalman filtering, RNN unrolling) but does not provide any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain any explicit statements about releasing the source code for their proposed method or provide a link to a code repository. |
| Open Datasets | Yes | For this, we use the publicly available datasets electricity and traffic [28]. ... This includes monthly and quarterly time series from the tourism competition dataset [2] describing tourism demand, hourly time series from the M4 competition [20] and parts dataset [6] which contains monthly demand of spare parts at a US auto-mobile company. |
| Dataset Splits | No | We train each method on all time series of these datasets but vary the size of the training range Ti {14, 21, 28} days. We evaluate all the methods on the next τ = 7 days after the forecast start time using the standard p50 and p90quantile losses. (The paper defines training and prediction ranges but does not explicitly mention a separate validation split.) |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments. It only mentions using the Amazon Sagemaker machine learning platform for one of the baselines (Deep AR) without further hardware specifications. |
| Software Dependencies | No | The paper mentions using a 'neural network framework (MXNet)' and 'R’s forecast package' for baselines, but it does not provide specific version numbers for these or any other software dependencies. |
| Experiment Setup | Yes | For Deep State the size of SSM (i.e., latent dimension) directly depends on the granularity of the time series which determines the number of seasons. For hourly data, we use hour-of-day (24 seasons) as well as day-of-week (7 seasons) models and hence latent dimension is 31. ... We train each method on all time series of these datasets but vary the size of the training range Ti {14, 21, 28} days. |