lendle

# Figure 1 algorithm from http://icml2008.cs.helsinki.fi/papers/361.pdf
# project a real vector onto a simplex in nlogn time
# v is the vector to project
# z is the value to which the projected vector must sum, one by default
function projectsimplex{T <: Real}(v::Array{T, 1}, z::T=one(T))
	n=length(v)
	µ = sort(v, rev=true)
	#finding ρ could be improved to avoid so much temp memory allocation
	ρ = maximum((1:n)[µ - (cumsum(µ) .- z) ./ (1:n) .> zero(T)]) 
	θ = (sum(µ[1:ρ]) - z)/ρ

lendle

lendle

lendle

hdps_screen <- function(outcome, treatment, covars, keep=0, indexes=TRUE) {
  #outcome and treatment are binary vectors of length n
  #covars is a matrix of binary variabesl of size n x p, 
  # where p is the number of covariates for in a paritular data dimension
  #keep: number of covariates to keep
  # indexes: return indexes or subest of covars
  #
  #returns indexes of kept covariates or matrix of kept covars
  #indexes are sorted by abs(log(multiplicitive bias))
  #if keep is 0 or greatner than the number of covars for which multiplicitive bias

lendle

library(Rcpp)
cppFunction('NumericVector colVars(NumericMatrix x) {
      int nrow = x.nrow(), ncol = x.ncol();
      NumericVector out(ncol);
      
      for (int j = 0; j < ncol; j++) {
        double mean = 0;
        double M2 = 0;
        int n;
        double delta, xx;

lendle

This is a markdown file

```jlcon other stuff on this line

julia> 1+3 4

julia> error("oops") ERROR: oops in error at error.jl:21

lendle

lendle

make_caret_wrappers <- function(alg_names) {
  
#creates an environment that stors SL wrappers.  For a given <name> in alg_names, the wrapper will be
# named SL.caret.<name>
# alg_names should be a list of names of models provided by the caret pacakge.
# see http://caret.r-forge.r-project.org/modelList.html

# only set up for binomial family
# also, set up to use single split CV internally within caret. This can be changed by fiddling with fitControl, below
  

lendle

import Base.convert

bitstype 64 IBMFloat64

# ibm float format is hex based:
#  * bit 1 = sign
#  * bits 2-8 = exponent to the power of 16, with bias of 64
#  * bits 9-64 = mantissa: no implied leading 1 (unlike ieee),
#    typically normalised (hex of bits 9-12 is non-zero), though this need
#    not be the case

lendle

module firmod

type FIR{in_t,coef_t}
    in::Vector{in_t}
    coef::Vector{coef_t}
    function FIR (coef::Vector{coef_t})
        new (zeros (in_t, size (coef)), copy (coef))
    end
end