tmastny

library(tidytext)
library(ggplot2)
library(dplyr)

diamonds %>%
  filter(cut %in% c("Fair", "Good")) %>%
  count(price = ifelse(price < 2500, "lo", "hi"), cut, clarity) %>%
  mutate(clarity = reorder_within(clarity, n, cut)) %>%
  ggplot(aes(clarity, n)) +
  geom_col() +

tmastny

# also works with sublime (subl)
code $(git diff --name-only master)

# relative is also useful depending on your working directory
code $(git diff --relative --name-only master) 

tmastny

library(ggplot2)
library(tidyr)
library(dplyr, warn.conflicts = FALSE)

diamonds_sum <- diamonds %>%
  mutate(across(cut, as.character)) %>%
  group_by(cut) %>%
  summarise(across(price, sum), across(carat, n_distinct), n = n())

diamonds_sum

tmastny

library(rlang)
"select"(iris, Species)
fn_var <- "select"

expr((!!sym(fn_var))(iris, Species) %>% head) %>%
  rlang::eval_tidy()

tmastny

library(recipes)
library(embed)
library(dplyr)
library(mlbench)

set.seed(124)

data(PimaIndiansDiabetes)

d <- PimaIndiansDiabetes

tmastny

library(tidyr)
library(dplyr)

set.seed(324)
d <- tibble(
  org = c(1, 1, 1, 2, 2, 3, 3, 3),
  sport = c("fb", "sc", "bb", "fb", "bb", "fb", "sc", "bb"),
  v1 = rnorm(8),
  v2 = rnorm(8) * 2,
  v3 = rnorm(8) * 10

tmastny

library(tidyverse)
test <- crossing(
  teamid = c(1, 2, 3),
  action = c("a", "b", "c")
)

test <- test %>%
  group_by(action) %>%
  mutate(
    day30 = c(10, 20, 30),

tmastny

library(caret)
library(dplyr)


threshold_method <- function(method) {
  thresh_code <- getModelInfo(method, regex = FALSE)[[1]]
  thresh_code$type <- c("Classification")

tmastny

# https://unix.stackexchange.com/a/402398/286677
"\e[A": history-search-backward
"\e[B": history-search-forward

tmastny

# http://sappingattention.blogspot.com/2018/06/meaning-chains-with-word-embeddings.html

cpath = function(matrix, w1, w2, blacklist = c()) {
p1 = matrix[[w1]]
p2 = matrix[[w2]]

base_similarity = cosineSimilarity(p1, p2)
pairsims = matrix %*% t(rbind(p1, p2))
# Words closer to *either* candidate than the other word. These might be used eventually; mostly to save on computation in later steps.
decentsims = pairsims[apply(pairsims, 1, max) >= base_similarity[1],]