Multivariate Aviation Time Series Modeling: VARs vs. LSTMs
نویسندگان
چکیده
Multivariate time-series modeling and forecasting constitutes an important problem with numerous applications. In this work, we consider multivariate continuous time series modeling from aviation, where the data consists of multiple sensor measurements from real world flights. While traditional approaches such as VAR (vector auto-regressive) models have been widely used for aviation time series, recent years have seen significant advances in modeling sequences using LSTMs (long short term memory models). In this paper, we do a careful empirical comparison between VAR and two types of LSTMs on multivariate aviation time series. Surprisingly, VAR is seen to significantly outperform LSTMs on real flight data, as well as synthetic data generated from VAR models and LSTM models. The results suggest that VAR is more suitable for multivariate continuous data especially from aviation where there may not be much utility for modeling long term memory which is a strength of LSTMs.
منابع مشابه
R2N2: Residual Recurrent Neural Networks for Multivariate Time Series Forecasting
Multivariate time-series modeling and forecasting is an important problem with numerous applications. Traditional approaches such as VAR (vector auto-regressive) models and more recent approaches such as RNNs (recurrent neural networks) are indispensable tools in modeling time-series data. In many multivariate time series modeling problems, there is usually a significant linear dependency compo...
متن کاملBayesian Multivariate Time Series Methods for Empirical Macroeconomics
Macroeconomic practitioners frequently work with multivariate time series models such as VARs, factor augmented VARs as well as timevarying parameter versions of these models (including variants with multivariate stochastic volatility). These models have a large number of parameters and, thus, over-parameterization problems may arise. Bayesian methods have become increasingly popular as a way o...
متن کاملExtremal financial risk models and portfolio evaluation
It is difficult to find an existing single model which is able to simultaneously model exceedances over thresholds in multivariate financial time series. A new modeling approach, which is a combination of max-stable processes, GARCH processes, and Markov processes, is proposed. Combining Markov processes and max-stable processes defines a new statistical model which has the flexibility of model...
متن کاملMultivariate time series modeling and classification via hierarchical VAR mixtures
A novel class of models for multivariate time series is presented. We consider hierarchical mixture-of-expert (HME) models in which the experts, or building blocks of the model, are vector autoregressions (VAR). It is assumed that the VAR-HME model partitions the covariate space, specifically including time as a covariate, into overlapping regions called overlays. In each overlay a given number...
متن کاملLearning to Diagnose with LSTM Recurrent Neural Networks
Clinical medical data, especially in the intensive care unit (ICU), consist of multivariate time series of observations. For each patient visit (or episode), sensor data and lab test results are recorded in the patient’s Electronic Health Record (EHR). While potentially containing a wealth of insights, the data is difficult to mine effectively, owing to varying length, irregular sampling and mi...
متن کامل