Online resources (R routines, case studies, etc)

This webpage proposes downloadable materials related to methods dealing with high dimensional data. It is very useful for practitionners wishing implementing themselves these methods.

Density-valued time series: Nonparametric density-on-density regression

• F. Ferraty, H. L. Shang. Density-valued time series: Nonparametric density-on-density regression (submitted). arXiv available online
  
  A comprehensive online supplementary material, which includes R implementation of the newly proposed methods is available at Scalar-on-function local linear regression and beyond
  • F. Ferraty, S. Nagy (2022). Scalar-on-function local linear regression and beyond. Biometrika, 109, 439-455 (available online)
    
    A comprehensive online supplementary material, which includes efficient R implementation of the newly proposed methods is available at https://bitbucket.org/StanislavNagy/fllr. It also provides additional simulations, hypotheses with discussion, proofs of the theorems, and some further theoretical considerations.
  
  

  ---

  Nested-kernel estimator

  • Ferraty F., Zullo A., Fauvel, M. (2019). Nonparametric regression on contaminated functional predictor with application to hyperspectral data. Econometrics and Statistics, 9, 95-107.
    
    The files given just below gathers all what you need for implementing this method:
    • Readme file (plain text)
    • R routines R routines source code containing all R routines dealing with nested-kernel estimator (plain text)
    • Implementation R commandlines allowing to implement the NKE methodology with different simulated datasets covering discrimination (i.e. responses = labels) as well as regression (scalar responses) setting (plain text)
  
  

  ---

  NONPARAMETRIC VARIABLE SELECTION (NOVAS)

  • Ferraty F., Hall P. (2015). An Algorithm for Nonlinear, Nonparametric Model Choice and Prediction. J. Comput. Graph. Stat., 24, 695-714 (arXiv:1401.8097).
    
    The files given just below gathers all what you need for implementing this method:
    • Readme file (plain text)
    • R routines R routines source code containing all R routines dealing with NOVAS (NOnparametric VAriable Selection) (plain text)
    • Case studies Useful file giving the R commandlines explaining how implementing in a very easy way this nonparametric variable selection. In particular, it allows to implement this method on different datasets dealing with food industry, petroleum or genomics problems (plain text)
    • Simulations R commandlines allowing to reproduce the simulation study leading to the results given in Table 4 of the above-mentioned work (plain text)
    • R package "Novas" This portable R package "Novas" (compressed archive file "Novas_1.0.tar.gz"), based on the R routines mentioned above, is available thanks to the work of Alex Fisch.
  
  

  ---

  NONPARAMETRIC REGRESSION WHEN BOTH RESPONSE AND PREDICTOR ARE RANDOM FUNCTIONS

  • Ferraty F., Van Keilegom I., Vieu, P. (2012). Regression when both Response and Predictor are Functions. J. Multivariate Anal., 109, 10-28.
    
    The files given just below gathers all what you need for implementing this method:
    • Readme file (plain text)
    • R routines R routines source code containing all R routines dealing with NPFDA (NonParametric Functional Data Analysis) ; it includes routines necessary for implementing the nonparametric regression when both response and predictor are functions (plain text)
    • Case studies Useful file giving the R commandlines explaining how implementing in a very easy way this functional nonparametric regression. In particular, it allows to reproduce all methodologies presented in the above-mentioned paper: estimations, predictions, simulations, asymptotic distributions, bootstrapped errors, pseudo-confidence area (plain text)
  
  

  ---

  MOST PREDICTIVE DESIGN POINTS (mpdp) FOR FUNCTIONAL DATA PREDICTORS

  • Ferraty F., Hall P., Vieu, P. (2010). Most predictive design points for functional data predictors. Biometrika, 97, 807-824.
    
    The files given just below gathers all what you need for implementing this method:
    • Readme file (plain text)
    • R "mpdp" routines R routines source code for implementing the stepwise algorithm (forward selection + backward deletion) allowing to select the most predictive design points (plain text)
    • Case studies Useful file giving the R commandlines explaining how implementing in a very easy way this stepwise algorithm on simulations and real datasets (plain text)
    • R "flr" routines Contains R routines of the alternative functional linear regression in order to make comparison with our stepwise algorithm (plain text)
  
  

  ---

  NONPARAMETRIC FUNCTIONAL DATA ANALYSIS (NPFDA)

  • 
    Click-on here to join the companion website of the book NonParametric Functional Data Analysis (Ferraty and Vieu, 2006 - Springer New York). This website provides R routines, cases studies, examples of use, data, etc.

  ---

  Copyright: All these available online materials are achieved, provided and maintained by Frederic Ferraty; they may be downloaded freely for your own personal study. Use for any commercial purpose is forbidden.