Identification of Structural VAR Models Via Independent Component Analysis: A Performance Evaluation Study

Moneta, Alessio; Pallante, Gianluca

doi:10.2139/ssrn.4109830

Cited by 4 publications

(4 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the past, several ICA-based algorithms have been developed. However, as shown by Moneta et al (2020), previous models are equivalent in terms of performance. Regarding VAR-DirectLiNGAM, it was proposed in order to solve the possible convergence issues of ICA-based methods (Himberg et al 2004) and it is guaranteed to retrieve the right solution of the problem if the model assumptions are satisfied and the sample size is very large.…”

Section: Comparison With Linear Non-gaussian Methodsmentioning

confidence: 99%

Multiscale Causal Structure Learning

D'Acunto¹,

Lorenzo²,

Barbarossa³

2022

Preprint

View full text Add to dashboard Cite

The inference of causal structures from observed data plays a key role in unveiling the underlying dynamics of the system. This paper exposes a novel method, named Multiscale-Causal Structure Learning (MS-CASTLE), to estimate the structure of linear causal relationships occurring at different time scales. Differently from existing approaches, MS-CASTLE takes explicitly into account instantaneous and lagged inter-relations between multiple time series, represented at different scales, hinging on stationary wavelet transform and non-convex optimization. MS-CASTLE incorporates, as a special case, a single-scale version named SS-CASTLE, which compares favorably in terms of computational efficiency, performance and robustness with respect to the state of the art onto synthetic data. We used MS-CASTLE to study the multiscale causal structure of the risk of 15 global equity markets, during covid-19 pandemic, illustrating how MS-CASTLE can extract meaningful information thanks to its multiscale analysis, outperforming SS-CASTLE. We found that the most persistent and strongest interactions occur at mid-term time resolutions. Moreover, we identified the stock markets that drive the risk during the considered period: Brazil, Canada and Italy. The proposed approach can be exploited by financial investors who, depending to their investment horizon, can manage the risk within equity portfolios from a causal perspective.

show abstract

Section: Comparison With Linear Non-gaussian Methodsmentioning

confidence: 99%

Multiscale Causal Structure Learning

D'Acunto¹,

Lorenzo²,

Barbarossa³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Fast independent component analysis is a type of ICA algorithm responsible for separating the unknown mixed signals to obtain useful independent signals using the source signal's independent and non-Gaussian nature [29]. An algorithm of FastICA works faster and is iteratively used at constant points with a simple structure and fast convergence [30].…”

Section: ) Fast Independent Component Analysismentioning

confidence: 99%

Emotional State Prediction Based on EEG Signals using Ensemble Methods

Alrebdi,

Al-Shargabi

2023

IJACSA

View full text Add to dashboard Cite

The emotional state is an essential factor that affects mental health. Electroencephalography (EEG) signal analysis is a promising method for detecting emotional states. Although multiple studies exist on EEG emotional signals classification, they have rarely considered processing time as a metric for classification model evaluation. Instead, they used either model accuracy and/or the number of features for evaluation. Processing time is an important factor to be considered in the context of mental health. Many people commonly use smart devices, such as smartwatches to monitor their emotional state and such devices require a short processing time. This research proposes an EEG-based model that detects emotional signals based on three factors: accuracy, number of features, and processing time. Two feature extraction algorithms were applied to EEG emotional signals: principal components analysis (PCA) and fast independent components analysis (FastICA). In the classification process, ensemble method classifiers were adopted due to their powerful performance. Three ensemble classifiers were used: random forest (RF), extreme gradient boosting (XGBoost), and adaptive boost (AdaBoost). The experimental results showed that the RF and XGBoost achieved the best accuracy, i.e., 95%, for both methods. However, XGBoost outperformed RF in terms of the number of features; it used 33 components extracted by PCA within 14 seconds, while RF used 36 within 4 seconds. AdaBoost was the worst in terms of both accuracy and processing time in the two experiments.

show abstract

“…ICA does not determine the sign nor the economic meaning of the shocks a priori. The columns of the (instantaneous) impact matrix should be reordered and if necessary their sign changed to make them easier to interpret economically (Gouriéroux et al, 2017;Moneta and Pallante, 2020).…”

Section: Identificationmentioning

confidence: 99%