Software

This site contains code for several purposes and its intended to work within R (www.r-project.org)

This computes Post Mortem Interval using Additive Models (AM) and Support Vector Machines (SVM) from the data obtained by José Ignacio Múñoz Barús and colleagues. The concentration of [K+], [Hx] and [Urea] in the vitreous humour is used to produce the estimation of the Post Mortem Interval.

This program was developed under windows but also woks in Unix with GUI=Tk. (R --gui=Tk).

Installation Instructions.

1. Download and install R from (www.r-project.org)
2. Install the necessary packages for CRAN: mgcv, kernlab
3. Download the following RData object: pmicalc.RData and save it to your default R folder.
4. Within a session with R, load this workspace with the command: load("your_folder/pmicalc.Rdata"). Some objects are added to your session:  eva2.txt, eval.txt, pmi, pmi.alpha, pmi.dvar, pmi.xvar, pmi.yvar, pmi.zvar, pmicalc, ptsinpolyg, resgam, resgam2, resgam2KU, resgamKU, ressvm, ressvmKU, tt
   
5. Type: require(kernlab); require(mgcv); require(tcltk); pmicalc()
6. Fill out the form and press the Button. The output includes the two graphics in separate windows and a text output with the prediction and confidence interval
7. pmi(K = 0, hxc = 0, Urea = 0, alpha = 0.05, pr = TRUE, plot = TRUE) do the same job but without using any Tcl/Tk command.

If you have any questions/comments, you can contact with  manuel.febrero@usc.es

The new R package fda.usc is avalaible through CRAN.

Simply, type in your R-session install.packages("fda.usc",dep=TRUE).

This package includes methods for :

1. Functional Data Representation
2. Exploratory Functional Data Analysis
3. Functional Outlier Detection
4. Functional Regression with Scalar Response
5. Functional Supervised and Non-Supervised Classification
6. Functional ANOVA

The purpose of this package is to complement the fda package by Jim Ramsay with own developments and those from the French group STAPH:(Working Group on Functional and Operator-based Statistics) in a integrated environment.

If you have any questions, comments or suggestions, contact with manuel.febrero@usc.es

geoR_NP.R contains some routines to estimate the variogram & ordinary kriging in a non parametric way. These routines follow  the style of library(geoR). This work is based on the following the papers:

1. GARCÍA-SOIDÁN, P.H.; FEBRERO-BANDE, M. and GONZÁLEZ-MANTEIGA, W. (2003). "Local linear regression estimation of the variogram". Statistics & Probability Letters Vol. 64, 169-179.
2. GARCÍA-SOIDÁN, P.H.; FEBRERO-BANDE, M. and GONZÁLEZ-MANTEIGA, W. (2004). "Nonparametric kernel estimation of an isotropic variogram.". Journal of Stat. Planning Inference. 121, 65-92.

Clearly, this routines are under development and are provided "as is" without any warranty. In future versions, I'll try to include some documentation.

Example Session in R

1. Put the file in your Working Directory.
2. Type source("geoR_NP.R")
3. The last four lines of this file contains an example using the s100 geodata example included in geoR.
   s100v.np<-varionp(s100) # Local Linear Regression Variogram
   s100v.SB<-varioShB(s100v.np) # Valid Shapiro-Botha approximation of a empirical or a nonparametric variogram
   s100k.np<-Kriging.NP(s100v.SB,s100) # Ordinary Kriging using Shapiro-Botha variogram

If you have any questions/comments, contact with  manuel.febrero@usc.es