Package {IFARIMA}

Type: Package
Title: Intuitionistic Fuzzy Autoregressive Integrated Moving Average Model
Version: 0.1.0
Description: The Intuitionistic Fuzzy Autoregressive Integrated Moving Average model offers advanced time series forecasting by integrating intuitionistic fuzzy sets with ARIMA. It enhances prediction accuracy by handling uncertainties and hesitations more effectively than traditional models, making it ideal for complex, real-world data applications. The package has been developed using the algorithm of Yolcu and Yolcu (2023) <doi:10.1016/j.eswa.2022.119336>.
License: GPL-3
Encoding: UTF-8
Imports: stats, forecast
Suggests: testthat (≥ 3.0.0)
Config/testthat/edition: 3
Config/roxygen2/version: 8.1.1
NeedsCompilation: no
Packaged: 2026-10-19 14:26:45 UTC; borap
Author: Anita Sarkar [aut], Ankit Kumar Singh [aut], Md Yeasin [aut, cre], Ranjit Kumar Paul [aut], Amrit Kumar Paul [aut], Pushkar Bora [aut]
Maintainer: Md Yeasin <yeasin.iasri@gmail.com>
Repository: CRAN
Date/Publication: 2026-10-29 11:50:09 UTC

Intuitionistic Fuzzy Autoregressive Integrated Moving Average Model

Description

Implements an intuitionistic fuzzy ARIMA forecasting model by integrating fuzzy clustering information with ARIMA modelling.

Usage

IFARIMA(Data, Class, Split)

Arguments

Data

Numeric time series data (a numeric vector without missing values).

Class

Number of fuzzy classes (at least 2).

Split

Proportion of the data used for training, strictly between 0 and 1.

Details

The membership matrix of the intuitionistic fuzzy clustering step is initialised at random, so results depend on the state of the random number generator. Call set.seed() before IFARIMA() to obtain reproducible results. The first round(length(Data) * Split) observations form the training set and the remainder the test set.

The membership degrees of an observation sum to one and the non-membership degrees are one minus the memberships, so some of these regressors are linearly dependent. Redundant columns are removed automatically before the ARIMA model with external regressors is fitted.

Value

A list with three components:

Accuracy

Accuracy measures (as returned by forecast::accuracy) for the training and test sets.

Train_Fitted

Numeric vector of fitted values on the training set.

Test_Prediction

Numeric vector of forecasts for the test set.

References

Yolcu, O.C. and Yolcu, U. (2023). A novel intuitionistic fuzzy time series prediction model with cascaded structure for financial time series. Expert Systems with Applications, 215, 119336. doi:10.1016/j.eswa.2022.119336

Examples

# Small toy example (runs in about a second)
set.seed(123)
dat <- 50 + cumsum(rnorm(60))

Model <- IFARIMA(Data = dat, Class = 2, Split = 0.8)
Model$Accuracy
head(Model$Train_Fitted)
Model$Test_Prediction

# Larger example with three clusters
set.seed(123)
dat2 <- rnorm(100, 60, 10)
Model2 <- IFARIMA(Data = dat2, Class = 3, Split = 0.8)
Model2$Accuracy