In systats/textlearnR:

knitr::opts_chunk$set(echo = T)

• taken from jlacko github

The underlying dataset consits of 9,000 tweets, collected on 2019-02-02 via rtweet. It consists of 1,500 tweets each from

• Hadley Wickham,
• Wes McKinney,
• François Chollet,
• Kim Kardashian,
• Kourtney Kardashian,
• Khloe Kardashian

From each account 1,200 tweets (80%) are included in training dataset and 300 (20%) in verification set.

pacman::p_load(tidyverse)

tweets <- read_csv("train_tweets.csv") %>% 
  mutate(split = T) %>% 
  bind_rows(read_csv("test_tweets.csv")) %>% 
  mutate(split = ifelse(is.na(split), F, split)) %>%
  glimpse

tweets %>% count(split)
#save(tweets, file = "tweets.Rdata")

celeb_mat <- tweets$name %>%
  dummies::dummy() 

celeb_names <- celeb_mat %>% 
  colnames %>% 
  str_remove("name|/Users/syro/MEGA/projects/celebrity-faceoff/code/keras_cnn.Rmd")

celeb_target <- celeb_mat %>% 
  as_tibble %>% 
  set_names(celeb_names) %>% 
  glimpse

set.seed(2019)
text_train <- tweets %>% filter(split) %>% pull(text)
text_test <- tweets %>% filter(!split) %>% pull(text)
y_train <- celeb_target[tweets$split, ] %>% as.matrix
y_test <- celeb_target[!tweets$split, ] %>% as.matrix

celeb_dat <- list(
  text_train = text_train, 
  text_test = text_test, 
  y_train = y_train, 
  y_test = y_test
)

celeb_dat %>% glimpse
save(celeb_dat, file = "celeb_dat.Rdata")

Old

• moved into the tuning process
• adds 2 new parameter

library(keras)
maxlen <- 60
max_features <- 13488

tokenizer <- text_tokenizer(num_words = max_features, lower = F, split = " ", char_level = F)
fit_text_tokenizer(tokenizer, train$text)

celeb$x_train <- tokenizer %>% 
  texts_to_sequences(train$text) %>%
  pad_sequences(maxlen = maxlen)

celeb$x_test <- tokenizer %>% 
  texts_to_sequences(test$text) %>%
  pad_sequences(maxlen = maxlen)

celeb %>% map(dim)
save(celeb, file = "celeb.Rdata")

systats/textlearnR documentation built on May 6, 2019, 8:31 p.m.