Social Media Analysis

This repository houses the code for social media analysis on Guyanese Facebook pages.

Prerequisites

Python 3.7.3 or greater. If you do not have Python installed, it is recommended to install Anaconda to get Python. Anaconda is an open-source package that includes Python and many popular Python packages. The current version of Anaconda is here

(Optional) A Git client. If you are on Windows, it is recommended to install Git Bash. Git Bash is a terminal application that includes Git and runs Linux commands on a Windows system. Git Bash can be found here

Setup

After installing Python, clone or download this repository. If cloning with Git, the command is:

git clone https://github.com/czig/nh19_sentiment.git

After cloning the repository, install the required Python packages, which are listed in requirements.txt file. An easy way to do this with the Anaconda package is to run the following command in a command prompt/terminal:

conda install --file requirements.txt

If not using Anaconda, use pip to install the packages:

pip install -r requirements.txt

Create the following directory structure at the root of the repository:

├── bigramCharts
├── lda_vis
├── models
├── numbers
├── tmp
├── topics
├── trigramCharts
├── wordclouds
└── wordFreqPlots

In order to pull data from Facebook, the facebookPull.py script needs your Facebook access token. To get your access token, log into your Facebook page and navigate to a public Facebook page. Right click on the page and select "View page source" from the drop down menu. A new tab should open, and once in the new tab, press Ctrl+F and search for "access_token". Move through the matches until you find a string that is 187 characters long and is a mix of letters and numbers. This string is your access token. Using a text editor, create a file named "access_token.txt" at the root of the directory. Place your access token, and only your access token, in this file and save. Now you will be able to pull Facebook posts and comments with the facebookPull.py script.

Usage

In order to view the usage for each script, run:

python <<filename>> -h

Accepted arguments and default values, along with a description of each argument, will be printed to the terminal. Arguments are handled with the argparse package, and each argument name is preceded by two hyphens and followed by a space. The value for the argument follows the space after the argument name. An example of running one of the scripts is below:

python generateAllWordCloud.py --type comments --pages guy --start_date 2019-01-01 --end_date 2019-07-29

A few of the scripts do not accept arguments, and thus, do not use the argparse package. These scripts can be run with python <<filename>>