Compară metode

	Predicția conformă pentru prognoza seriilor de timp	Modelul ARIMA (Autoregresiv Integrat cu Medii Mobile)	Gradient Boosting
Domeniu	Econometrie	Econometrie	Învățare automată
Familie	Regression model	Regression model	Machine learning
Anul apariției	2021	2015	2001
Autorul original	Angelopoulos & Bates (tutorial); Xu & Xie (time-series EnbPI)	Box & Jenkins (Box-Jenkins methodology)	Friedman, J. H.
Tip	Distribution-free prediction interval wrapper	Univariate time-series model	Ensemble (sequential boosting of decision trees)
Sursa seminală	Angelopoulos, A. N. & Bates, S. (2023). Conformal Prediction: A Gentle Introduction. Foundations and Trends in Machine Learning, 16(4), 494-591.	Box, G. E. P., Jenkins, G. M., Reinsel, G. C. & Ljung, G. M. (2015). Time Series Analysis: Forecasting and Control (5th ed.). Wiley. ISBN: 978-1118675021	Friedman, J. H. (2001). Greedy Function Approximation: A Gradient Boosting Machine. Annals of Statistics, 29(5), 1189–1232.
Denumiri alternative	conformal prediction, distribution-free prediction intervals, EnbPI, Konformal Tahmin (Conformal Prediction — Zaman Serisi)	Box-Jenkins model, ARIMA(p,d,q), ARIMA Modeli	Gradient Boosting (GBM), GBM, gradient boosted trees, gradient boosting machine
Înrudite	4	5	5
Rezumat	Conformal prediction is a distribution-free wrapper that turns any point forecaster — ARIMA, a neural network, or a machine-learning model — into valid prediction intervals using only its residuals. The time-series form was popularised by Xu & Xie (2021) and the modern tutorial treatment by Angelopoulos & Bates (2023).	ARIMA is a univariate time-series forecasting model that combines autoregressive, integrated (differencing), and moving-average components to predict a single continuous series from its own past. It is the centrepiece of the Box-Jenkins methodology set out in Box, Jenkins, Reinsel & Ljung's Time Series Analysis (5th ed., 2015).	Gradient Boosting is an ensemble learning method, formalised by Jerome H. Friedman in 2001, that combines a sequence of weak learners — typically shallow decision trees — so that each new tree is fitted to minimise the residual errors of the trees before it. It is the core algorithm behind popular implementations such as XGBoost, LightGBM and CatBoost.