Probabilistic Forecasting with Spline Quantile Function RNNs

By Jan Gasthaus, Konstantinos Benidis, Bernie Wang, Syama Sundar Rangapuram, David Salinas, Valentin Flunkert, Tim Januschowski. 2019

201 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processes (2019, methodology): “A concurrent approach was proposed in [35] which also uses Copula and estimates marginal quantile functions with the approach proposed in [11] and models the Cholesky factor as the output of a neural network.”
Deep Factors for Forecasting (2019, background): “…in extracting complex patterns from large datasets of related time series by exploiting similar smoothness and temporal dynamics, and common responses to exogenous input, i.e. dependencies of type (1) and (3) (Flunkert et al., 2017; Wen et al., 2017; Mukherjee et al., 2018; Gasthaus et al., 2019).”
Forecasting: theory and practice (2020, background): “Variations are possible, with either non-standard output distributions in forecasting such as the multinomial distribution (Rabanser et al., 2020) or via representing the probability density as cumulative distribution function (Salinas et al., 2019a) or the quantile function (Gasthaus et al., 2019).”
Recurrent Neural Networks for Time Series Forecasting: Current Status and Future Directions (2019, methodology): “Other examples for successful new developments in the field are novel architectures such as DeepAR, Multi-Quantile Recurrent Neural Network (MQRNN), Spline Quantile Function RNNs and Deep State Space Models for probabilistic forecasting (Flunkert et al., 2017; Wen et al., 2017; Gasthaus et al., 2019”
A Survey on Bayesian Deep Learning (2016, methodology): “In addition, the signal passing through the RNN decays exponentially, making it impossible to model long-term dependencies in long sequences.”
GluonTS: Probabilistic Time Series Models in Python (2019, methodology): “Table 1 shows the CRPS (Gasthaus et al., 2019) accuracy of different forecast methods in GluonTS on the following 11 public datasets:”
Probabilistic Forecasting with Temporal Convolutional Neural Network (2019, methodology): “More recently, Gasthaus et al. (2019) propose SQF-RNN, a probabilistic framework to model conditional quantile functions with isotonic splines, which allows more ﬂexible output distributions.”
Deep Learning for Time Series Forecasting: Tutorial and Literature Survey (2020, methodology): “SQF-RNN [64] RNN Prob 1 CRPS QL, MSIS, NRMSE, OWA demand, traffic, count data, finance, M4 Models non-parametric distributions with splines”