R: tidytext NASA data

library(jsonlite)
metadata <- fromJSON("https://data.nasa.gov/data.json")
names(metadata$dataset)
save(metadata,file="metadata.Rdata")

class(metadata$dataset$title)
class(metadata$dataset$description)
class(metadata$dataset$keyword)

# ambil judul2
library(dplyr)
nasa_title <- data_frame(id = metadata$dataset$`_id`$`$oid`,
                         title = metadata$dataset$title)
nasa_title
save(nasa_title,file="nasa_title.Rdata")

# sampling 5 line
nasa_desc <- data_frame(id = metadata$dataset$`_id`$`$oid`,
                        desc = metadata$dataset$description)
nasa_desc %>%
   select(desc) %>%
      sample_n(5)

# cari keyword
library(tidyr)
nasa_keyword <- data_frame(id = metadata$dataset$`_id`$`$oid`,
                           keyword = metadata$dataset$keyword) %>%
                                 unnest(keyword)
nasa_keyword
save(nasa_desc,file="nasa_desc.Rdata")
save(nasa_keyword,file="nasa_keyword.Rdata")

library(tidytext)
nasa_title <- nasa_title %>%
    unnest_tokens(word, title) %>%
       anti_join(stop_words)
nasa_desc <- nasa_desc %>%
    unnest_tokens(word, desc) %>%
       anti_join(stop_words)
nasa_title
nasa_desc

# count & sort
nasa_title %>%
    count(word, sort = TRUE)
nasa_desc %>%
    count(word, sort = TRUE)

# buang keyword yang tidak ada artinya
my_stopwords <- data_frame(word = c(as.character(1:10),
                           "v1", "v03", "l2", "l3", "l4", "v5.2.0",
                           "v003", "v004", "v005", "v006", "v7"))
nasa_title <- nasa_title %>%
     anti_join(my_stopwords)
nasa_desc <- nasa_desc %>%
     anti_join(my_stopwords)

# supaya OCEAN & Ocean jadi sama
nasa_keyword <- nasa_keyword %>%
     mutate(keyword = toupper(keyword))

#
library(widyr)
title_word_pairs <- nasa_title %>%
     pairwise_count(word, id, sort = TRUE, upper = FALSE)
title_word_pairs

# ini bisa gagal karena terlalu besar
desc_word_pairs <- nasa_desc %>%
    pairwise_count(word, id, sort = TRUE, upper = FALSE)
desc_word_pairs

library(ggplot2)
library(igraph)
library(ggraph)
set.seed(1234)
title_word_pairs %>%
   filter(n >= 250) %>%
       graph_from_data_frame() %>%
           ggraph(layout = "fr") +
               geom_edge_link(aes(edge_alpha = n, edge_width = n), edge_colour = "cyan4") +
               geom_node_point(size = 5) +
               geom_node_text(aes(label = name), repel = TRUE,
               point.padding = unit(0.2, "lines")) +
               theme_void()

Pranala Menarik

• R