In Guscode/Sentida: Sentiment Scoring Of Words And Sentences

knitr::opts_chunk$set(echo = TRUE)

#Loading this will enable you to work with æøå - VERY IMPORTANT
Sys.setlocale(category = "LC_ALL", locale = "UTF-8")

#load packages and data
library(pacman)
p_load(ggplot2,stringr,dplyr,tidyverse, remotes) #add your favorite packages

fv_all <- read_csv("fv_all.csv")

#look at the data, check if classes make sense, e.g. $text should be character
head(fv_all$text, 10)
colnames()
class()

#Load sentida and apply it to one of the tweets, try both total and mean
remotes::install_github("Guscode/Sentida", force = T)
library(Sentida)
?sentida


#select a leader from one of the Danish political parties (https://www.dr.dk/nyheder/politik/resultater/folketingsvalg). From fv_all make a new dataframe including only tweets mentioning this person. 
#E.g.: paludan <- filter(fv_all, str_detect(fvhash$text, "Paludan") == T)



### You might want to do some preprocessing of the text here. str_remove_all() is cute, tolower() is cool, str_replace_all() is sexy. Make the tweets lowercase and remove all numbers (Major hint alert: str_remove_all(paludan$text,[:digit:]))



#Make a new row in the dataframe with sentiment score of every tweet mentioning this person (e.g use lapply() or make a loop if you dare)
?lapply

#what is his/her mean sentiment score?
###Bonus: What is their mean sentiment score per day? !WARNING ADVANCED PLEASE SKIP! lifesaver: aggregate()




#Compare your political leader to one from another party





#Did the parties who won less mandates than the last election have a worse sentiment score than the rest?




#Are there any correlations between sentiment score and amount of favourites/retweets?

Guscode/Sentida documentation built on Dec. 29, 2021, 4:38 a.m.