Rcpp for DataFrame access

Rcpp-DataFrame.R

library(Rcpp)

cppFunction('
int countSteps (DataFrame df, double thresh=0.001) {
  
  int out = 0;
  if(df.nrows()==0) return out;
  
  int last_i = 0;
  out++;
  
  NumericVector xs = df[0];
  NumericVector ys = df[1];
  
  for(int i=1; i<xs.length(); i++) {
    if(fabs(xs[i]-xs[last_i])>thresh ||
       fabs(ys[i]-ys[last_i])>thresh) {
      out++;
      last_i = i;
    }
  }
  
  return out;
}')


cppFunction('
int countSteps2 (DataFrame df, double thresh=0.001) {

  int out = 0;
  int n = df.nrows();
  if (n==0) return out;

  int last_i = 0;
  out++;

  int m = df.length();

  for(int i=1; i<n; i++) {
    for(int j=0; j<m; j++) {
      const NumericVector& col_j = df[j];
      if(fabs(col_j[i]-col_j[last_i])>thresh) {
        out++;
        last_i = i;
        break;
      }
    }
  }

  return out;
}')

cppFunction('
int countSteps3 (NumericMatrix nm, double thresh=0.001) {

  int out = 0;
  int n = nm.nrow();
  if (n==0) return out;

  int last_i = 0;
  out++;

  int m = nm.ncol();

  for(int i=1; i<n; i++) {
    for(int j=0; j<m; j++) {
      NumericMatrix::Column col_j = nm(_,j);
      if(fabs(col_j[i]-col_j[last_i])>thresh) {
        out++;
        last_i = i;
        break;
      }
    }
  }

  return out;
}')


cppFunction('
int countSteps4 (DataFrame df, double thresh=0.001) {
  
  int out = 0;
  int n = df.nrows();
  if (n==0) return out;
  
  int last_i = 0;
  out++;

  int m = df.length();
  
  NumericVector cols[m];
  for(int j=0; j<m; j++) {
    cols[j] = df[j];
  }

  for(int i=1; i<n; i++) {
    for(int j=0; j<m; j++) {
      if(fabs(cols[j][i]-cols[j][last_i])>thresh) {
        out++;
        last_i = i;
        break;
      }
    }
  }
  
  return out;
}')

df <- data.frame(x=cumsum(rnorm(2e7)), y=cumsum(rnorm(2e7)))

print(system.time(x <- countSteps(df,3)))
print(system.time(x2 <- countSteps2(df,3)))
print(system.time(x3 <- countSteps3(data.matrix(df),3)))
print(system.time(x4 <- countSteps4(df,3)))