CODEWITHSUNDEEP
r
Adam_G
Adam_G

Reputation: 7879

Function call in mapply returning list not data frame

I have a custom function. When I run the function manually, it returns a data frame:

> create_sentiment_df('taggreason', 'republican', 'lost')
Joining, by = "word"
Joining, by = "word"
      sentiment prop.sentiment twitter.name      party election.result
1         anger     0.04721931   taggreason republican            lost
2  anticipation     0.14375656   taggreason republican            lost
3       disgust     0.01259182   taggreason republican            lost
4          fear     0.06190976   taggreason republican            lost
5           joy     0.09024134   taggreason republican            lost
6      negative     0.10073452   taggreason republican            lost
7      positive     0.26862539   taggreason republican            lost
8       sadness     0.03777545   taggreason republican            lost
9      surprise     0.03882476   taggreason republican            lost
10        trust     0.19832109   taggreason republican            lost

However, I want to run this multiple times, so I am using mapply on each row of a data frame. Here is the data, with a data frame of only 1 row (for testing):

> datt1
# A tibble: 1 x 4
  twtr_handle party      result district_flipped
  <chr>       <chr>      <chr>  <chr>           
1 taggreason  republican lost   flipped

And then the function call:

rslt <- mapply(create_sentiment_df, datt1$twtr_handle, datt1$party, datt1$result)

Which returns:

> rslt
                taggreason  
sentiment       Character,10
prop.sentiment  Numeric,10  
twitter.name    Character,10
party           Character,10
election.result Character,10

or:

enter image description here

Below is the function. It requires twitter authorization, so I'm not sure how it can easily be rerun. Is there anything about the function itself that would make mapply return a list rather than a data frame?

library(rtweet)
library(tidytext)
library(tidyverse)
library(BBmisc)
library(reshape)

create_token(
  app = "Flippable Sentiment Analysis",
  consumer_key = c_k,
  consumer_secret = c_s,
  access_token <- a_t,
  access_secret <- a_s)

create_sentiment_df <- function(twitter.name, party, election.result) {
  va_stop_words <- stop_words %>% select(-lexicon) %>% 
    bind_rows(data.frame(word = c("https", "t.co", "rt", "amp")))
  nrc_lex <- get_sentiments("nrc") # many sentiments

  dat <- get_timeline(twitter.name, n=3200)
  dat$created_at <- as.Date(dat$created_at)
  dat_2017 <- subset(dat, created_at > as.Date('2017-01-01') & created_at < as.Date('2017-11-06'))

  dat_words <- dat_2017 %>% 
    select(status_id, text) %>% 
    unnest_tokens(word,text)

  dat_words_interesting <- dat_words %>% anti_join(va_stop_words)
  dat_sentiment <- dat_words_interesting %>% left_join(nrc_lex)
  dat_sentiment_count <- dat_sentiment %>% 
    filter(!is.na(sentiment)) %>% 
    group_by(sentiment) %>% 
    summarise(prop.sentiment=n())
  dat_sentiment_count <- na.omit(dat_sentiment_count)

  dat_sentiment_count <- cbind(dat_sentiment_count[1], 
                            prop.table(data.matrix(dat_sentiment_count[-1]), margin=2))

  # dat_sentiment_count$twitter.name <- NA
  dat_sentiment_count$twitter.name <- twitter.name
  dat_sentiment_count$party <- party
  dat_sentiment_count$election.result <- election.result

  return(as.data.frame(dat_sentiment_count))
}

Upvotes: 1

Views: 300

Answers (1)

Mankind_2000
Mankind_2000

Reputation: 2208

Your create_sentiment_df function returns a data.frame and mapply simplifies it by default.

If you need the list of data.frames, you can do:

mapply(create_sentiment_df, datt1$twtr_handle, datt1$party, datt1$result, SIMPLIFY = FALSE)

If you require a single data.frame for all your data.frame outputs, use:

do.call(rbind, mapply(create_sentiment_df, datt1$twtr_handle,
                      datt1$party, datt1$result, SIMPLIFY = FALSE))

Upvotes: 1

Related Questions