Forecasting Models

31 forecasting algorithms from exponential smoothing to intermittent demand.

Categories

Exponential Smoothing

Weighted average methods

ETSAutoETSHolt-Winters+5 more

ARIMA

Autoregressive models

ARIMAAutoARIMA

Theta Methods

Decomposition-based forecasting

ThetaAutoThetaDynamicTheta+2 more

Multi-Seasonal

Complex seasonality patterns

TBATSMSTLMFLES+3 more

Baseline Models

Simple benchmark methods

NaiveSeasonalNaiveSMA+1 more

Intermittent Demand

Sparse/lumpy demand patterns

CrostonADIDATSB+3 more

All Models

All CategoryARIMABaselineExponential SmoothingIntermittentMulti-SeasonalTheta

Model	Category	Description
ETS	Exponential Smoothing	Error-Trend-Seasonality state space
AutoETS	Exponential Smoothing	Automatic ETS selection
SES	Exponential Smoothing	Simple Exponential Smoothing
SESOptimized	Exponential Smoothing	Optimized SES parameters
Holt	Exponential Smoothing	Linear trend method
HoltWinters	Exponential Smoothing	Trend + seasonality
SeasonalES	Exponential Smoothing	Seasonal exponential smoothing
SeasonalESOptimized	Exponential Smoothing	Optimized seasonal ES
ARIMA	ARIMA	AutoRegressive Integrated Moving Average
AutoARIMA	ARIMA	Automatic ARIMA (p,d,q) selection
Theta	Theta	Theta method for short-term forecasts
AutoTheta	Theta	Automatic Theta method
OptimizedTheta	Theta	Optimized Theta parameters
DynamicTheta	Theta	Dynamic Theta method
DynamicOptimizedTheta	Theta	Dynamic + optimized Theta
TBATS	Multi-Seasonal	Trigonometric seasonality, Box-Cox, ARMA
AutoTBATS	Multi-Seasonal	Automatic TBATS selection
MSTL	Multi-Seasonal	Multiple Seasonal-Trend decomposition
AutoMSTL	Multi-Seasonal	Automatic MSTL configuration
MFLES	Multi-Seasonal	Median-based Feature-Logic Expert System
AutoMFLES	Multi-Seasonal	Automatic MFLES configuration
Naive	Baseline	Repeat last observation
SeasonalNaive	Baseline	Repeat last seasonal period
SMA	Baseline	Simple Moving Average
RandomWalkDrift	Baseline	Random walk with drift
CrostonClassic	Intermittent	Classic Croston method
CrostonOptimized	Intermittent	Optimized Croston parameters
CrostonSBA	Intermittent	Syntetos-Boylan Approximation
ADIDA	Intermittent	Aggregate-Disaggregate method
IMAPA	Intermittent	Intermittent Multiple Aggregation
TSB	Intermittent	Teunter-Syntetos-Babai method

Showing 31 of 31

Model Selection Guide

Data Type	Recommended	Alternative
Clean, trending	AutoETS	Theta
Seasonal	AutoETS	HoltWinters
Multiple seasonal	AutoTBATS	MSTL
Noisy data	AutoMFLES	AutoARIMA
Sparse demand	CrostonClassic	ADIDA
Short-term	Theta	SeasonalNaive
Long-term	AutoARIMA	AutoTBATS
Limited data	SeasonalNaive	Theta