artemklevtsov’s gists

artemklevtsov / bin.R

Last active April 9, 2018 13:32

Binning (discretize) variables based on CART

	library(recipes)
	library(dplyr)
	library(tibble)

	step_bin <- function(recipe, ..., role = "predictor", trained = FALSE,
	threshold = .05, n.group = NULL, woe = FALSE, objects = NULL,
	skip = FALSE) {
	if (threshold <= 0)
	stop("`threshold` should be greater than zero", call. = FALSE)
	if (threshold >= 1)

artemklevtsov / Telco-Customer-Churn.R

Last active April 9, 2018 13:02

Telco-Customer-Churn

	## ---- Загрузка пакетов ----

	library(readr) # чтение данных
	library(dplyr) # манипулации с данными
	library(tidyr) # манипулации с данными
	library(rsample) # разбиение выборки
	library(recipes) # препроцессинг
	library(yardstick) # метрики модели
	library(glmnet) # логистическая регрессия с регуляризацией

artemklevtsov / cramer.R

Last active March 1, 2018 14:22

Статистика Крамера

	library(data.table)

	#' @title Функция для расчёта статистики Крамера
	#' @param x Категориальная переменная.
	#' @param y Категориальная переменная.
	#' @return Статистика Крамера (число в диапазоне от 0 до 1).
	cramer <- function(x, y) {
	# На случай сравнения переменной с самой собой
	if (identical(x, y)) return(1.0)
	# Наблюдаемые частоты

artemklevtsov / hero_stats.R

Last active September 8, 2017 17:44

	library(curl)
	library(jsonlite)
	library(data.table)

	# Get data
	hero_stats <- stream_in(curl("https://api.opendota.com/api/heroStats"), verbose = FALSE)
	# Post processing
	setDT(hero_stats)
	hero_roels <- dcast(
	data = res[, .(roles = unlist(roles)), by = id],

artemklevtsov / svn.R

Created July 14, 2017 14:20

Get R releases dates

	library(xml2)
	library(curl)
	library(data.table)
	library(lubridate)

	r_svn_url <- "https://svn.r-project.org/R/tags/"
	h <- handle_setheaders(new_handle(customrequest = "PROPFIND"), Depth="1")
	req <- curl_fetch_memory(r_svn_url, handle = h)
	doc <- read_xml(rawToChar(req$content))
	ns <- xml_ns(doc)

artemklevtsov / df2lst.cpp

Last active April 9, 2017 06:34

	#include <Rcpp.h>
	using namespace Rcpp;

	RObject get_elem(const RObject& x, std::size_t i) {
	switch(x.sexp_type()) {
	case INTSXP: {
	RObject res = wrap(INTEGER(x)[i]);
	if (Rf_isFactor(x)) {
	res.attr("class") = "factor";
	res.attr("levels") = x.attr("levels");

artemklevtsov / hash.cpp

Last active February 5, 2017 05:54

	// [[Rcpp::plugins("cpp11")]]

	#include <Rcpp.h>
	#include <functional>
	using namespace Rcpp;

	// [[Rcpp::export]]
	std::vector<size_t> hash(const std::vector<std::string>& x) {
	std::vector<size_t> res(x.size());
	std::transform(x.begin(), x.end(), res.begin(), std::hash<std::string>());

artemklevtsov / hostname.cpp

Last active January 13, 2017 18:42

	// [[Rcpp::plugins("cpp11")]]
	// [[Rcpp::depends("BH")]]

	#include <Rcpp.h>
	#include <boost/asio.hpp>

	using namespace Rcpp;
	using namespace boost::asio;
	using namespace boost::asio::ip;

artemklevtsov / flatten.cpp

Last active March 14, 2017 15:44

	// [[Rcpp::plugins("cpp11")]]

	#include <Rcpp.h>
	using namespace Rcpp;

	void append_string(std::vector<std::string>& s1, const std::string s2) {
	if (!s2.empty()) {
	std::transform(s1.begin(), s1.end(), s1.begin(),
	[&] (std::string& s) { return s2 + '.' + s; });
	}

artemklevtsov / plot-roc.R

Last active February 10, 2017 18:16

Artem Klevtsov artemklevtsov