Thuật ngữ AI
Từ điển đầy đủ về Trí tuệ nhân tạo
Page-Hinkley Method
Online change detection test based on calculating the cumulative sum of differences between observations and their mean, with an automatic reset mechanism.
Structural Change
Fundamental modification in the underlying mechanism generating the data, affecting model parameters or relationships between variables.
Changing Variance
Type of change detected when the volatility or dispersion of observations varies significantly between different periods of the time series.
Changing Mean
Change characterized by a modification of the average level of the time series, often indicating a shift in the underlying generating process.
Seasonality Detection
Identification of changes in the cyclical or periodic patterns of a time series, including the emergence, disappearance, or modification of seasonal patterns.
ADF Test
Augmented statistical test checking for the presence of a unit root in a time series, essential for detecting stationarity changes.
Bayesian Changepoint Method
Probabilistic approach using Bayesian inference to estimate the posterior distribution of change points and quantify the associated uncertainty.
E-Divisive Algorithm
Non-parametric detection method based on recursive division of the series into segments maximizing a divergence statistic between sub-series.
PELT
Algorithme d'optimisation exact en temps linéaire avec élagage, permettant de trouver efficacement les points de changement multiples en minimisant un critère de coût.
Détecteur de Winters
Méthode spécialisée dans la détection de changements dans les modèles de lissage exponentiel de Holt-Winters, particulièrement sensible aux variations saisonnières.