Generalized Adaptive Capped Estimator for Time Series Forecasting

Provides deterministic forecasting for weekly, monthly, quarterly, and yearly time series using the Generalized Adaptive Capped Estimator. The method includes preprocessing for missing and extreme values, extraction of multiple growth components (including long-term, short-term, rolling, and drift-based signals), volatility-aware asymmetric capping, optional seasonal adjustment via damped and normalized seasonal factors, and a recursive forecast formulation with moderated growth. The package includes a user-facing forecasting interface and a plotting helper for visualization. Related forecasting background is discussed in Hyndman and Athanasopoulos (2021) < https://otexts.com/fpp3/> and Hyndman and Khandakar (2008) . The method extends classical extrapolative forecasting approaches and is suited for operational and business planning contexts where stability and interpretability are important.

GACE

Generalized Adaptive Capped Estimator
A stable, deterministic forecasting engine for weekly, monthly, quarterly, and yearly data.

Overview

GACE provides a transparent, tuning-free forecasting method based on hybrid growth signals and adaptive asymmetric caps.
It extends deterministic capped-growth forecasting to support weekly, monthly, quarterly, and yearly time series with optional seasonal scaling.

The method is designed for:

• demand forecasting
• operational forecasting
• financial and portfolio forecasting
• budgeting and scenario planning

The philosophy: Stable + Interpretable + Fast
No nonlinear optimization. No stochastic fitting. Fully deterministic.

Features

• Supports weekly / monthly / quarterly / yearly data
• Hybrid growth signals (year-over-year, short-term, rolling, drift)
• Volatility-aware asymmetric caps
• Seasonal and non-seasonal modes
• Deterministic, extremely fast computation
• plot_gace() for visualization
• Lightweight dependencies

Installation

Install the released version from CRAN:

install.packages("GACE")

Example

library(GACE)

set.seed(1)
y <- ts(rnorm(60, mean = 100, sd = 10), frequency = 12)

fc <- gace_forecast(y, periods = 12, freq = "month")
plot_gace(fc)