On the use of cross-validation for time series predictor evaluation

By Christoph Bergmeir, José Manuel Benítez. Information Sciences, 2012

1,121 citations recorded by the source corpus.

Subfields: Management Science and Operations Research, Signal Processing

View Paper: https://www.semanticscholar.org/paper/f93f1a82d37ce0e6eeb421c89a441258bc7cf329

How researchers see this paper

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

In Management Science and Operations Research

In Signal Processing

Citations in the corpus

A hybrid method for imputation of missing values using optimized fuzzy c-means with support vector regression and a genetic algorithm (2013, background): “If the missing values are not treated correctly, they may decrease or even jeopardize the validity of the research [3,5,14,22,34].”
Enhancing quantitative intra-day stock return prediction by integrating both market news and stock prices information (2014, methodology): “When applying crossvalidation on time series data, special care has to be taken [1,2].”
A note on the validity of cross-validation for evaluating autoregressive time series prediction (2018, methodology): “In addition, the serial correlation in the data, along with possible non-stationarities, make the use of CV appear problematic as it does not account for these issues (Bergmeir and Benı́tez, 2012).”
Evaluating time series forecasting models: an empirical study on performance estimation methods (2019, methodology): “.f. Figure 8 for a sample graph. For the rst two cases, S1 and S2, real-valued roots of the characteristic polynomial are sampled from the uniform distribution [ r; 1:1] [[1:1;r], where r is set to 5 [4]. Afterwards, the roots are used to estimate the models and create the time series. The data is t”
A Hierarchy of Limitations in Machine Learning (2020, background): “”67 In a series of simulations, Bergmeir et al. (2018) consider different cross-validation schemes on an autoregressive series.”
Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure (2017, neutral): “When models are intended to predict in time, blocks may be drawn in the same manner (i.e. blocks of contiguous time) to better ensure independence between cross-validation folds (Burman et al. 1994, Racine 2000, Bergmeir and Benitez 2012).”