Table 5 Common operation differentiating anger/non-anger across participants
From: Towards PPG-based anger detection for emotion regulation
Operation family | # Sessions | Description |
|---|---|---|
MF GARCH_ar_P1_Q2 | 11 | Fits a Generalized AutoRegressive Conditional Heteroskedasticity (GARCH) model to the time-series (order P = 1, Q = 2). Explores the appropriateness of model |
IN_AutoMutualInfoStats_diff_20_gaussian | 11 | Automutual information statistics on the differences of the time-series. Uses gaussian estimation with a max delay of 20 |
MF_arfit_1_8_sbc | 10 | Fits Autoregressive (AR) models from order P = 1 to 8 on the time series. Optimal model is selected with Schwartz’s Bayesian Criterion (SBC). Statistics on model coefficients, final prediction error, and eigendecomposition, etc |
SB_MotifThree_diffquant | 8 | Coarse grain motifs of an equiprobable three level alphabet (ABC) on the time-series differences. Outputs proportion of motifs ranging from word lengths 1 to 4 |
MF_ExpSmoothing_05_best | 8 | Fits an exponential smoothing model, by using half of the time-series as a training set to find the optimal smoothing parameter: alpha. Outputs fitting parameters and statistics on residuals |
MF_AR_arcov_5 | 7 | Fits an AR model of order 5 to the time series. Outputs parameters of model and residual analysis |
MF_StateSpace_n4sid | 7 | Fits a state space model to the time series. Trains on first half of the time-series and predicts on second half. Outputs model parameters and statistics on residuals |
SP_Summaries_fft | 7 | Power spectrum statistics using Fast Fourier transform (e.g., peaks, bandwidth, shape of cumulative sum, etc.) |
CO_Embed2_Basic_tau | 7 | Properties of a point density embedding in 2D space (e.g., output of points near diagonals and geometric shapes) |
WL_fBM | 6 | Wavelet estimation of fractional Brownian motion or Gaussian noise in the time series |