UCSD Confessions Analysis

Author: Curtis Lee, Meihan Zhang, Tyler Farnan Summary: Hello! This repository contains a jupyter notebook demonstration of exploratory sentiment analysis on UCSD Confessions, an anomyous online forum for students.

Dataset Origin

UCSD Confessions Public Archive

Third Party Modules Used

Numpy, Pandas, Matplotlib, Seaborn, Glob, NLTK, TextBlob, Scikit-learn, Gensim, Wordcloud

Requirements for Sentiment Analysis:

To run the sentiment analysis portion, extra features are needed:

$ python -m textblob.download_corpora

Then run python and download the data set:

>>> import nltk
>>> nltk.download('stopwords')

File Stucture

This repo is broken down into multiple folders.

• code contains all the necessary python modules
  • some of the code modules may use each other
• data contains all the dataset files used in this project
  • xlsx files contain the UCSD confessions data
  • academic_calendar.json contains breakdown of the UCSD Quarter System
  • _corpus.txt files contains list of words used for keyword trend analysis
  • .pkl and .dat cache files generated one time for fast access later
  • other files are generated by the lexicon analysis process
• output is a dedicated folder for the graphs generated by the process

Note: some of the code modules have demo features and can be run independantly, but must in the root directory of repo

for example: $ python load_data.py

How to run code:

1. Clone the repo
2. Install all the necessary third party modules and dependancies
3. Please see main.ipynb for full usage details