Analysis of Twitter Data on the 2019 Indian Parliamentary Elections by Hemanth Bharatha Chakravarthy
This project aims to investigate scraped twitter data on the ongoing 2019 Indian Lok Sabha Elections and perform sentiment analysis, forecasting, and analysis on bot tweets based on this data.
Link: https://hemanth-bharatha-chakravarthy.shinyapps.io/lok_sabha/
What words were most used? Were they favorable or unfavorable? What were the broad sentiments on the BJP's campaign and how did these sentiments evolve over time? Built using sentiment analysis and text mining.
Tracking down bots and stopping their misinformation requires understanding them--this analyses their performance over time in the day. This is built to be interactive.
This project is written in R and it's UI is built with RShiny to be interactive and reactive. I use rtweet to scrape twitter data (more on rtweet here: https://rtweet.info/) and use syuzhet to perform sentiment analysis (more on syuzhet here: https://www.rdocumentation.org/packages/syuzhet/versions/1.0.4/topics/get_nrc_sentiment). Aside of some geospatial analysis and mapping tools, a large part of the core functionality of this project is built in dplyr and ggplot2.
The csv file: https://drive.google.com/file/d/1cSUnWV-dQK0XgBfJGkhHJNX_6uc1JDgK/view?usp=sharing
This data is a mixed sample of the most popular and most recent one hundred and eight thousand tweets made in English from the last three days (as of 20 March, 2019, 10:07 pm EST) on the Bharatiya Janata Party's #MainBhiChowkidar campaign.
There is one main file that powers the app: app.R in /lok_sabha/. (https://github.com/b-hemanth/lok_sabha_public/lok_sabha/app.R) There are two other R script files in this same folder, namely: helpers.R and static.R. helpers.R contains much of the preprocessing for the app. It reads in the data, cleans it, does a lot of the text-mining, and does some preliminary sentiment analysis. It pushes out some .Rds files and dataframes that are then used in static.R to produce the static images. These are those plots that do not employ reactive variables and are not interactive on the final interface. Finally, ../lok_sabha/static/ is the folder that contains the static images produced in the R script files. These static images are then rendered in the Shiny App.
Note: IF RUNNING THE STATIC, you must switch the datafile in helpers.R to work out of the right directory. This is because when the Shiny App is run, it assumes that the folder of the app.R file is the working directory whereas the regular R script assumes that the upper root directory is the pwd.