SAMoSSA: Multivariate Singular Spectrum Analysis with Stochastic Autoregressive Noise
Authors: Abdullah Alomar, Munther Dahleh, Sean Mann, Devavrat Shah
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Through representative empirical studies, we validate the superior performance of SAMo SSA compared to existing baselines. Notably, SAMo SSA s ability to account for AR noise structure yields improvements ranging from 5% to 37% across various benchmark datasets. |
| Researcher Affiliation | Academia | Abdullah Alomar MIT aalomar@mit.edu; Munther Dahleh MIT dahleh@mit.edu; Sean Mann MIT seanmann@mit.edu; Devavrat Shah MIT devavrat@mit.edu |
| Pseudocode | Yes | 3 Algorithm. The proposed algorithm provides two main functionalities... For a visual depiction of the algorithm, refer to Figure 1. |
| Open Source Code | No | The paper does not provide an explicit statement or link for the open-source code of the methodology described. |
| Open Datasets | Yes | Traffic Dataset. This public dataset obtained from the UCI repository shows the occupancy rate of traffic lanes in San Francisco... Electricity Dataset. This is a public dataset obtained from the UCI repository which shows the 15-minutes electricity load of 370 households... Exchange Dataset. This is a dataset containing the daily exchange rates of eight foreign currencies... |
| Dataset Splits | Yes | Each dataset was split into train, validation, and test sets (see Appendix B.1). Specifically, in our testing period, we do 1-hour ahead forecasts for the next 48 hours. ... Traffic Dataset... use the first 10248 time-points for training, the next 48 points for validation, and another 48 points for testing... |
| Hardware Specification | No | The paper does not provide specific details regarding the hardware used for the experiments. |
| Software Dependencies | No | The paper mentions software like 'Prophet s Python library', 'Python library statsmodels', 'Keras implementation', and 'Gluon TS package', but it does not specify any version numbers for these software dependencies. |
| Experiment Setup | Yes | The relevant hyperparameters are as follows: 1. The number of retained singular values, k. ... 2. The shape parameter of the stacked Page matrix... 3. The number of autoregressive lag coefficients, p. ... Prophet. ...Changepoint prior scale. ... Seasonality prior scale. ... Seasonality Mode. ... ARIMA. ... Autoregressive order. ... Differencing order. ... Moving average order. ... LSTM. ... number of layers {2, 3, 4}. ... Deep AR. We use the default parameters. |