Ironholds

if(k < 2){

    drink_to_forget()

}

Ironholds

SELECT regexp_extract(uri_path, '/((\\d+px-)?[^/]+)$', 1) as uri_file
FROM wmf_raw.webrequest
WHERE year = 2014 AND month = 09 AND day = 23 AND
hour = 07 AND webrequest_source='upload' LIMIT 5;

Ironholds

//R
min       lq      mean
200422 201139.5 202649.46
    15     18.0     33.01
median     uq    max neval
201427 201824 258190   100
    37     51     56   100

//C++
min    lq     mean median

Ironholds

//Includes/namespaces
#include <Rcpp.h>
using namespace Rcpp;

// Recursive session counting function
// [[Rcpp::export]]
int session_count(NumericVector x, int local_minimum) {
  
  //Create output object
  int count;

Ironholds

intertimes <- function(x, local_minimum){

return(length(x[x > local_minimum]) + 1)

}

Ironholds

 test   Freq
1     Cable 416932
2 Corporate  13848
3   Dial-up  77302
4   Invalid  17842
5   Unknown  71175

Ironholds

#Intertime computation timings, for a test vector of 3.4k elements.

#Original C++ implementation 
expr   min   lq     mean median    uq    max neval
 {     test <- intertime(test_vec) } 3.982 4.25 7.805621 4.3405 4.564 687.78  1000

#C++ implementation with RCpp's "diff()" - syntactic sugar around adjacent_difference
                                 expr   min    lq     mean median    uq max neval
 {     test <- intertime2(test_vec) } 4.181 4.354 9.668984  4.437 4.696 823  1000

Ironholds

> z <- c(TRUE, TRUE, FALSE, FALSE, TRUE, FALSE, TRUE, TRUE, TRUE)
> 
> rle(z)
Run Length Encoding
  lengths: int [1:5] 2 2 1 1 3
  values : logi [1:5] TRUE FALSE TRUE FALSE TRUE

Ironholds

#For our test object, we have x, a data.table
timestamp                tz order
 1: 20140907142430     Europe/Warsaw     1
 2: 20141005162245      Europe/Paris     2
 3: 20140908142859       Europe/Rome     3
 4: 20140903093741      Asia/Yerevan     4
 5: 20140805102142 America/Sao_Paulo     5
 6: 20140905142222   Europe/Brussels     6
 7: 20140819225906       Europe/Rome     7
 8: 20140811190410     Europe/London     8

Ironholds

  x <- x[, j = {
    
    #Copy object
    copy.dt <- copy(.SD)
    
    #Convert timestamps
    timestamps <- mw_strptime(copy.dt$timestamp)
    
    #Extract timestamps, convert as appropriate.
    copy.dt$original_timestamp <- as.character(timestamps)