adaptiveFTS

---

Description

adaptiveFTS provides tools for adaptive estimation procedures for weakly dependent functional time series. Developed by Maissoro et al. (2024, 2025), it includes:

• Estimators for local regularity parameters
• Mean function estimation
• Autocovariance function estimation
• An adaptive Best Linear Unbiased Predictor (BLUP)

This package leverages the computational efficiency of Rcpp and RcppArmadillo for high-performance functional data analysis.

---

Table of Contents

1. Features
2. Installation
3. Usage
4. Examples
5. Contributing
6. License

---

Features

• Adaptive estimators for functional time series analysis.
• Efficient computation using Rcpp and RcppArmadillo.
• Implements methodologies from cutting-edge research (2024, 2025).
• Tools for predicting functional time series via BLUP.

---

Installation

From GitHub

To install the development version of adaptiveFTS, use the devtools package:

# Install devtools if not already installed
install.packages("devtools")

# Install adaptiveFTS from GitHub
devtools::install_github("hmaissoro/adaptiveFTS")

---

Usage

Load the package and start using its functions:

# Load adaptiveFTS
library(adaptiveFTS)

# Example of mean function estimation

## Load data
data("data_far")

## Estimate risk function for the mean
dt_mean_risk <- estimate_mean_risk(
  data = data_far, idcol = "id_curve", tcol = "tobs", ycol = "X",
  t = c(1/4, 1/2, 3/4), bw_grid = NULL,
  kernel_name = "epanechnikov"
)

(Refer to the examples section for detailed code snippets and use cases.)

---

Examples

Example of mean function estimation

# Load data
data("data_far")

# Estimate risk function for the mean
dt_mean_risk <- estimate_mean_risk(
  data = data_far, idcol = "id_curve", tcol = "tobs", ycol = "X",
  t = c(1/4, 1/2, 3/4), bw_grid = NULL,
  kernel_name = "epanechnikov"
)

# Visualize mean risk at various observation points
dt_dcast <- data.table::dcast(data = dt_mean_risk, formula = h ~ t, value.var = "mean_risk")
manipulateWidget::combineWidgets(
  list = list(
    dygraphs::dygraph(
      data = dt_dcast[, list(h, "t = 0.25" = `0.25`)],
      main = "t = 0.25", xlab = "h", ylab = "Risk Function"),
    dygraphs::dygraph(
      data = dt_dcast[, list(h, "t = 0.5" = `0.5`)],
      main = "t = 0.5", xlab = "h", ylab = "Risk Function"),
    dygraphs::dygraph(
      data = dt_dcast[, list(h, "t = 0.75" = `0.75`)],
      main = "t = 0.75", xlab = "h", ylab = "Risk Function")
  ),
  nrow = 3
)

# Estimate mean function with optimal bandwidths
dt_mean <- estimate_mean(
  data = data_far, idcol = "id_curve", tcol = "tobs", ycol = "X",
  t = c(1/4, 1/2, 3/4), bw_grid = seq(0.005, 0.15, len = 45),
  kernel_name = "epanechnikov"
)

# Display rounded estimates of the mean function
DT::datatable(data = dt_mean[, lapply(.SD, function(X) round(X, 3))])

---

Contributing

We welcome contributions! Follow these steps:

1. Fork the repository: adaptiveFTS
2. Clone your fork:

   git clone https://github.com/your-username/adaptiveFTS.git

3. Create a feature branch:

   git checkout -b feature-name

4. Commit your changes:

   git commit -m "Add feature description"

5. Push the branch:

   git push origin feature-name

6. Submit a pull request.

---

License

This package is licensed under the AGPL (>= 3) license. See the LICENSE file for more details.

---

Acknowledgements

• Built using Rcpp and RcppArmadillo for efficient computation.
• Inspired by methodologies developed in Maissoro et al. (2024, 2025).

---

References

1. Hassan Maissoro, Valentin Patilea, and Myriam Vimond. Adaptive Estimation for Weakly Dependent Functional Time Series. 2024. Available at arXiv:2403.13706.

2. Hassan Maissoro, Valentin Patilea, and Myriam Vimond. Adaptive prediction for Functional Times Series. 2024. Work in progress.

---