npregfast: 'npregfast': Nonparametric Estimation of Regression Models...

In npregfast: Nonparametric Estimation of Regression Models with Factor-by-Curve Interactions

npregfast: Nonparametric Estimation of Regression Models with Factor-by-Curve Interactions.

Description

This package provides a method for obtain nonparametric estimates of regression models using local polynomial kernel smoothers or splines. Particular features of the package are facilities for fast smoothness estimation, and the calculation of their first and second derivative. Users can define the smoothers parameters. Confidence intervals calculation is provided by bootstrap methods. Binning techniques were applied to speed up computation in the estimation and testing processes.

Details

npregfast is designed along lines similar to those of other R regression packages. The main function of the library is frfast which, by default, fits a nonparametric regression model based on local polynomial kernel smoothers. Note that through the argument formula users can decide to fit a model by taking or not taking the interaction into account and by the argument formula it is posible to select the type of smoother: kernel or splines. Numerical and graphical summaries of the fitted object can be obtained by using the generic functions, print.frfast, summary.frfast and plot.frfast. Another of these generic functions is predict.frfast, which takes a fitted model of the frfast class and, given a new data set of values of the covariate, produces predictions. As mentioned above, this package can be used to fit models taking into account factor-by-curve interactions. In this framework, it will be necessary to ascertain if the factor produces an effect on the response and thus, there is a interaction or, in contrast, the estimated regression curves are equal. To this end, the package provides the globaltest function which answers this question through a bootstrap-based test. If the factor results significant, then plotdiff() enables the user to obtain a graphical representation that shows the differences between the estimated curves (estimate, first or second derivative) for any set of two levels of the factor. Additionally, with critical() it is possible to obtain the value of the covariate that maximises the estimate and first derivative of the function and the value of the covariate that equals the second derivative to zero, for each of these levels. Again, to test if these estimated points are equal for all levels, the package provides the localtest function. Note that, to compare these points between any set of two levels, a confidence interval for the difference can be obtained by applying criticaldiff().

For a listing of all routines in the NPRegfast package type: library(help="npregfast").

View a demo Shiny app or see the full README on GitHub.

Author(s)

Marta Sestelo, Nora M. Villanueva and Javier Roca-Pardinas.

References

Efron, B. (1979). Bootstrap methods: another look at the jackknife. Annals of Statistics, 7, 1–26.

Efron, E. and Tibshirani, R. J. (1993). An introduction to the Bootstrap. Chapman and Hall, London.

Huxley, J. S. (1924). Constant differential growth-ratios and their significance. Nature, 114:895–896.

Sestelo, M. (2013). Development and computational implementation of estimation and inference methods in flexible regression models. Applications in Biology, Engineering and Environment. PhD Thesis, Department of Statistics and O.R. University of Vigo.

Sestelo, M. and Roca-Pardinas, J. (2011). A new approach to estimation of length-weight relationship of Pollicipes pollicipes (Gmelin, 1789) on the Atlantic coast of Galicia (Northwest Spain): some aspects of its biology and management. Journal of Shellfish Research, 30(3), 939–948.

Sestelo, M., Villanueva, N.M., Meira-Machado, L., Roca-Pardinas, J. (2017). npregfast: An R Package for Nonparametric Estimation and Inference in Life Sciences. Journal of Statistical Software, 82(12), 1-27.

Wand, M. P. and Jones, M. C. (1995). Kernel Smoothing. Chapman & Hall, London.

npregfast documentation built on Nov. 5, 2025, 6:17 p.m.